You are on page 1of 426


Continuous change in protective relaying has been caused by two different influences. One is the fact that the
requirements imposed by power systems are in a constant state of change, and our understanding of the basic
concepts has sharpened considerably over the years. The other is that the means of implementing the fundamental
concepts of fault location and removal and system restoration are constantly growing more sophisticated.
It is primarily because of these changing constraints that this text has been revised and expanded. It began with
contributions from two giants of the industry, J. Lewis Blackburn and George D. Rockefeller. From the nucleus of
their extensive analyses and writings, and the desire to cover each new contingency with new relaying concepts, this
volume has evolved. New solutions to age-old problems have become apparent as greater experience has been
gained. No problem is without benefit in the solution found.
This new edition weeds out those relaying concepts that have run their course and have been replaced by more
perceptive methods of implementation using new solid-state or microprocessor-based devices.
No single technological breakthrough has been more influential in generating change than the microprocessor.
Initially, the methods of translating a collection of instantaneous samples of sine waves into useful current,
direction, and impedance measurements were not obvious. Diligent analysis and extensive testing allowed these
useful functions to be obtained and to be applied to the desired protective functions. This text attempts to describe,
in the simplest possible terms, the manner in which these digital measurements are accomplished in present-day
In addition to those already mentioned, huge contributions were made in the development and refinement of the
concepts described in this book by Hung Jen Li, Walter Hinman, Roger Ray, James Crockett, Herb Lensner, Al
Regotti, Fernando Calero, Eric Udren, James Greene, Liancheng Wang, Elmo Price, Solveig Ward, John
McGowan, and Cliff Downs. Some of these names may not be immediately recognizable, but all have made an
impact with their thoughtful, accurate, well-reasoned writings, and they all deserve the gratitude of the industry for
the wealth of knowledge they have contributed to this book. I am keenly aware of the high quality of the technical
offerings of these people, and I am particularly grateful for the warmth and depth of their friendship.

Walter A. Elmore


Preface iii

1 Introduction and General Philosophies 1

Revised by W. A. Elmore
1 Introduction 1
2 Classification of Relays 1
2.1 Analog/Digital/Numerical 2
3 Protective Relaying Systems and Their Design 2
3.1 Design Criteria 3
3.2 Factors Influencing Relay Performance 4
3.3 Zones of Protection 4
4 Applying Protective Relays 4
4.1 System Configuration 5
4.2 Existing System Protection and Procedures 5
4.3 Degree of Protection Required 5
4.4 Fault Study 5
4.5 Maximum Loads, Transformer Data, and Impedances 6
5 Relays and Application Data 6
5.1 Switchboard Relays 6
5.2 Rack-Mounted Relays 7
6 Circuit-Breaker Control 8
7 Comparison of Symbols 9

2 Technical Tools of the Relay Engineer: Phasors, Polarity, and Symmetrical Components 11
Revised by W. A. Elmore
1 Introduction 11
2 Phasors 11
2.1 Circuit Diagram Notation for Current and Flux 11
2.2 Circuit Diagram Notation for Voltage 12

vi Contents

2.3 Phasor Notation 12

2.4 Phasor Diagram Notation 13
2.5 Phase Rotation vs. Phasor Rotation 15
3 Polarity in Relay Circuits 15
3.1 Polarity of Transformers 15
3.2 Polarity of Protective Relays 15
3.3 Characteristics of Directional Relays 16
3.4 Connections of Directional Units to Three-Phase Power Systems 17
4 Faults on Power Systems 18
4.1 Fault Types and Causes 18
4.2 Characteristics of Faults 20
5 Symmetrical Components 21
5.1 Basic Concepts 21
5.2 System Neutral 23
5.3 Sequences in a Three-Phase Power System 23
5.4 Sequence Impedances 24
5.5 Sequence Networks 26
5.6 Sequence Network Connections and Voltages 27
5.7 Network Connections for Fault and General Unbalances 28
5.8 Sequence Network Reduction 29
5.9 Example of Fault Calculation on a Loop-Type Power System 32
5.10 Phase Shifts Through Transformer Banks 37
5.11 Fault Evaluations 39
6 Symmetrical Components and Relaying 42

3 Basic Relay Units 43

Revised by W. A. Elmore
1 Introduction 43
2 Electromechanical Units 43
2.1 Magnetic Attraction Units 43
2.2 Magnetic Induction Units 45
2.3 D’Arsonval Units 47
2.4 Thermal Units 47
3 Sequence Networks 47
3.1 Zero Sequence Networks 47
3.2 Composite Sequence Current Networks 48
3.3 Sequence Voltage Networks 49
4 Solid-State Units 50
4.1 Semiconductor Components 50
4.2 Solid-State Logic Units 52
4.3 Principal Logic Units 52
5 Basic Logic Circuits 54
5.1 Fault-Sensing Data Processing Units 54
5.2 Amplification Units 59
5.3 Auxiliary Units 59
6 Integrated Circuits 63
6.1 Operational Amplifier 63
6.2 Basic Operational Amplifier Units 65
6.3 Relay Applications of Operational Amplifier 68
7 Microprocessor Architecture 70
Contents vii

4 Protection Against Transients and Surges 71

W. A. Elmore
1 Introduction 71
1.1 Electrostatic Induction 71
1.2 Electromagnetic Induction 72
1.3 Differential- and Common-Mode Classifications 72
2 Transients Originating in the High-Voltage System 73
2.1 Capacitor Switching 73
2.2 Bus Deenergization 73
2.3 Transmission Line Switching 74
2.4 Coupling Capacitor Voltage Transformer (CCVT) Switching 74
2.5 Other Transient Sources 74
3 Transients Originating in the Low-Voltage System 74
3.1 Direct Current Coil Interruption 74
3.2 Direct Current Circuit Energization 75
3.3 Current Transformer Saturation 75
3.4 Grounding of Battery Circuit 75
4 Protective Measures 75
4.1 Separation 75
4.2 Suppression at the Source 77
4.3 Suppression by Shielding 77
4.4 Suppression by Twisting 77
4.5 Radial Routing of Control Cables 78
4.6 Buffers 78
4.7 Optical Isolators 78
4.8 Increased Energy Requirement 79

5 Instrument Transformers for Relaying 81

W. A. Elmore
1 Introduction 81
2 Current Transformers 81
2.1 Saturation 81
2.2 Effect of dc Component 82
3 Equivalent Circuit 82
4 Estimation of Current Transformer Performance 82
4.1 Formula Method 83
4.2 Excitation Curve Method 83
4.3 ANSI Standard: Current Transformer Accuracy Classes 85
5 European Practice 87
5.1 TPX 88
5.2 TPY 88
5.3 TPZ 88
6 Direct Current Saturation 88
7 Residual Flux 89
8 MOCT 91
9 Voltage Transformers and Coupling Capacitance Voltage Transformers 91
9.1 Equivalent Circuit of a Voltage Transformer 91
9.2 Coupling Capacitor Voltage Transformers 92
9.3 MOVT/EOVT 93
10 Neutral Inversion 93
viii Contents

6 Microprocessor Relaying Fundamentals 95

W. A. Elmore
1 Introduction 95
2 Sampling Problems 97
3 Aliasing 97
4 How to Overcome Aliasing 98
4.1 Antialiasing Filters 98
4.2 Nonsynchronous Sampling 98
5 Choice of Measurement Principle 99
5.1 rms Calculation 100
5.2 Digital Filters 100
5.3 Fourier-Notch Filter 100
5.4 Another Digital Filter 101
5.5 dc Offset Compensation 101
5.6 Symmetrical Component Filter 102
5.7 Leading-Phase Identification 102
5.8 Fault Detectors 102
6 Self-Testing 103
6.1 Dead-Man Timer 103
6.2 Analog Test 103
6.3 Check-Sum 103
6.4 RAM Test 103
6.5 Nonvolatile Memory Test 103
7 Conclusions 104

7 System Grounding and Protective Relaying 105

Revised by W. A. Elmore
1 Introduction 105
2 Ungrounded Systems 105
2.1 Ground Faults on Ungrounded Systems 105
2.2 Ground Fault Detection on Ungrounded Systems 107
3 Reactance Grounding 108
3.1 High-Reactance Grounding 108
3.2 Resonant Grounding (Ground Fault Neutralizer) 109
3.3 Low-Reactance Grounding 109
4 Resistance Grounding 110
4.1 Low-Resistance Grounding 110
4.2 High-Resistance Grounding 111
5 Sensitive Ground Relaying 112
5.1 Ground Overcurrent Relay with Conventional Current Transformers 112
5.2 Ground Product Relay with Conventional Current Transformers 113
5.3 Ground Overcurrent Relay with Zero Sequence Current Transformers 114
6 Ground Fault Protection for Three-Phase, Four-Wire Systems 114
6.1 Unigrounded Four-Wire Systems 114
6.2 Multigrounded Four-Wire Systems 115

8 Generator Protection 117

Revised by C. L. Downs
1 Introduction 117
2 Choice of Technology 117
Contents ix

3 Phase Fault Detection 117

3.1 Percentage Differential Relays (Device 87) 118
3.2 High Impedance Differential Relays (Device 87) 119
3.3 Machine Connections 119
3.4 Split-Phase 119
4 Stator Ground Fault Protection 120
4.1 Unit-Connected Schemes 120
4.2 95% Ground Relays 120
4.3 Neutral-to-Ground Fault Detection (Device 87N3) 121
4.4 100% Winding Protection 122
5 Backup Protection 123
5.1 Unbalanced Faults 123
5.2 Balanced Faults 124
6 Overload Protection 126
6.1 RTD Schemes (Device 49) 126
6.2 Thermal Replicas (Device 49) 126
7 Volts per Hertz Protection 126
8 Overspeed Protection 126
9 Loss-of-Excitation Protection 127
9.1 Causes of Machine Loss of Field 127
9.2 Hazard 127
9.3 Loss-of-Field Relays 128
9.4 KLF and KLF-1 Curves 129
9.5 Two-Zone KLF Scheme 129
10 Protection Against Generator Motoring 130
10.1 Steam Turbines 131
10.2 Diesel Engines 131
10.3 Gas Turbines 131
10.4 Hydraulic Turbines 131
11 Inadvertent Energization 132
12 Field Ground Detection 134
12.1 Brush-Type Machine 135
12.2 Brushless Machines 136
12.3 Injection Scheme for Field Ground Detection 136
13 Alternating-Current Overvoltage Protection for Hydroelectric Generators 136
14 Generator Protection at Reduced Frequencies 136
15 Off-Frequency Operation 138
16 Recommended Protection 139
17 Out-of-Step Protection 139
18 Bus Transfer Systems for Station Auxiliaries 139
18.1 Fast Transfer 139
18.2 Choice of Fast Transfer Scheme 140
18.3 Slow Transfer 142
19 Microprocessor-Based Generator Protection 143

9 Motor Protection 145

Revised by C. L. Downs
1 Introduction 145
1.1 General Requirements 145
1.2 Induction Motor Equivalent Circuit 146
1.3 Motor Thermal Capability Curves 146
x Contents

2 Phase-Fault Protection 147

3 Ground-Fault Protection 147
4 Locked-Rotor Protection 149
5 Overload Protection 153
6 Thermal Relays 153
6.1 RTD-Input-Type Relays 154
6.2 Thermal Replica Relays 154
7 Low-Voltage Protection 155
8 Phase-Rotation Protection 155
9 Negative Sequence Voltage Protection 155
10 Phase-Unbalance Protection 156
11 Negative Sequence Current Relays 157
12 Jam Protection 157
13 Load Loss Protection 157
14 Out-of-Step Protection 158
15 Loss of Excitation 158
16 Typical Application Combinations 159

10 Transformer and Reactor Protection 163

Revised by J. J. McGowan
1 Introduction 163
2 Magnetizing Inrush 163
2.1 Initial Inrush 163
2.2 Recovery Inrush 165
2.3 Sympathetic Inrush 165
3 Differential Relaying for Transformer Protection 166
3.1 Differential Relays for Transformer Protection 166
3.2 General Guidelines for Transformer Differential Relaying Application 171
4 Sample Checks for Applying Transformer Differential Relays 173
4.1 Checks for Two-Winding Banks 173
4.2 Checks for Multiwinding Banks 178
4.3 Modern Microprocessor Relay 180
5 Typical Application of Transformer Protection 180
5.1 Differential Scheme with Harmonic Restraint Relay Supervision 180
5.2 Ground Source on Delta Side 182
5.3 Three-Phase Banks of Single-Phase Units 183
5.4 Differential Protection of a Generator-Transformer Unit 183
5.5 Overexcitation Protection of a Generator-Transformer Unit 184
5.6 Sudden-Pressure Relay (SPR) 185
5.7 Overcurrent and Backup Protection 185
5.8 Distance Relaying for Backup Protection 192
5.9 Overcurrent Relay with HRU Supplement 192
6 Typical Protective Schemes for Industrial and Commercial Power Transformers 193
7 Remote Tripping of Transformer Bank 197
8 Protection of Phase-Angle Regulators and Voltage Regulators 197
9 Zig-Zag Transformer Protection 202
10 Protection of Shunt Reactors 203
10.1 Shunt Reactor Applications 203
10.2 Rate-of-Rise-of-Pressure Protection 205
10.3 Overcurrent Protection 205
10.4 Differential Protection 206
Contents xi

10.5 Reactors on Delta System 207

10.6 Turn-to-Turn Faults 209

11 Station-Bus Protection 213

Revised by Solveig Ward
1 Introduction 213
1.1 Current Transformer Saturation Problem and Its Solutions on Bus Protection 213
1.2 Information Required for the Preparation of a Bus Protective Scheme 215
1.3 Normal Practices on Bus Protection 215
2 Bus Differential Relaying with Overcurrent Relays 216
2.1 Overcurrent Differential Protection 216
2.2 Improved Overcurrent Differential Protection 216
3 Multirestraint Differential System 217
4 High Impedance Differential System 219
4.1 Factors that Relate to the Relay Setting 221
4.2 Factors that Relate to the High-Voltage Problem 221
4.3 Setting Example for the KAB Bus Protection 222
5 Differential Comparator Relays 222
6 Protecting a Bus that Includes a Transformer Bank 223
7 Protecting a Double-Bus Single-Breaker with Bus Tie Arrangement 224
8 Other Bus Protective Schemes 226
8.1 Partial Differential Relaying 226
8.2 Directional Comparison Relaying 227
8.3 Fault Bus (Ground-Fault Protection Only) 227

12 Line and Circuit Protection 229

Revised by Elmo Price
1 Introduction 229
1.1 Classification of Electric Power Lines 229
1.2 Techniques for Line Protection 229
1.3 Seleting a Protective System 229
1.4 Relays for Phase- and Ground-Fault Protection 230
1.5 Multiterminal and Tapped Lines and Weak Feed 230
2 Overcurrent Phase- and Ground-Fault Protection 231
2.1 Fault Detection 231
2.2 Time Overcurrent Protection 232
2.3 Instantaneous Overcurrent Protection 237
2.4 Overcurrent Ground-Fault Protection 238
3 Directional Overcurrent Phase- and Ground-Fault Protection 239
3.1 Criteria for Phase Directional Overcurrent Relay Applications 239
3.2 Criteria for Ground Directional Overcurrent Relay Applications 239
3.3 Directional Ground-Relay Polarization 239
3.4 Mutual Induction and Ground-Relay Directional Sensing 243
3.5 Applications of Negative Sequence Directional Units for Ground Relays 244
3.6 Selection of Directional Overcurrent Phase and Ground Relays 244
4 Distance Phase and Ground Protection 247
4.1 Fundamentals of Distance Relaying 247
4.2 Phase-Distance Relays 250
4.3 Ground-Distance Relays 254
4.4 Effect of Line Length 257
4.5 The Infeed Effect on Distance-Relay Application 260
xii Contents

4.6 The Outfeed Effect on Distance-Relay Applications 261

4.7 Effect of Tapped Transformer Bank on Relay Application 261
4.8 Distance Relays with Transformer Banks at the Terminal 262
4.9 Fault Resistance and Ground-Distance Relays 265
4.10 Zero Sequence Mutual Impedance and Ground-Distance Relays 265
5 Loop-System Protection 267
5.1 Single-Source Loop-Circuit Protection 267
5.2 Multiple-Source Loop Protection 269
6 Short-Line Protection 270
6.1 Definition of Short Line 270
6.2 Problem Associated with Short-Line Protection 270
6.3 Current-Only Scheme for Short-Line Protection 270
6.4 Distance Relay for Short-Line Protection 270
7 Series-Capacitor Compensated-Line Protection 273
7.1 A Series-Capacitor Compensated Line 273
7.2 Relaying Quantities Under Fault Conditions 273
7.3 Distance Protection Behavior 275
7.4 Practical Considerations 276
8 Distribution Feeder Protection 276
8.1 Relay Coordination with Reclosers and Sectionalizers on a Feeder 277
8.2 Coordinating with Low-Voltage Breaker and Fuse 277
Appendix A: Equation (12-2) 281
Appendix B: Impedance Unit Characteristics 281
B.1 Introduction 281
B.2 Basic Application Example of a Phase Comparator 284
B.3 Basic Application Example of a Magnitude Comparator 285
B.4 Practical Comparator Applications in Distance Relaying 285
B.5 Reverse Characteristics of an Impedance Unit 294
B.6 Response of Distance Units to Different Types of Faults 298
B.7 The Influence of Current Distribution Factors and Load Flow 302
B.8 Derived Characteristics 305
B.9 Apparent Impedance 305
B.10 Summary 306
Appendix C: Infeed Effect on Ground-Distance Relays 306
C.1 Infeed Effect on Type KDXG, LDAR, and MDAR Ground-Distance Relays 306
C.2 Infeed Effect on Type SDG and LDG Ground-Distance Relays 307
Appendix D: Coordination in Multiple-Loop Systems 308
D.1 System Information 308
D.2 Relay Type Selection 308
D.3 Relay Setting and Coordination 309

13 Backup Protection 323

Revised by E. D. Price
1 Introduction 323
2 Remote vs. Local Backup 323
2.1 Remote Backup 323
2.2 Local Backup and Breaker Failure 324
2.3 Applications Requiring Remote Backup with Breaker-Failure Protection 326
3 Breaker-Failure Relaying Applications 327
3.1 Single-Line/Single-Breaker Buses 327
3.2 Breaker-and-a-Half and Ring Buses 328
Contents xiii

4 Traditional Breaker-Failure Scheme 329

4.1 Timing Characteristics of the Traditional Breaker-Failure Scheme 329
4.2 Traditional Breaker-Failure Relay Characteristics 330
4.3 Microprocessor Relays 331
5 An Improved Breaker-Failure Scheme 332
5.1 Problems in the Traditional Breaker-Failure Scheme 332
5.2 The Improved Breaker-Failure Scheme 333
5.3 Type SBF-1 Relay 334
6 Open Conductor and Breaker Pole Disagreement Protection 336
7 Special Breaker-Failure Scheme for Single-Pole Trip-System Application 337
14 System Stability and Out-of-Step Relaying 339
W. A. Elmore
1 Introduction 339
2 Steady-State Stability 339
3 Transient Stability 340
4 Relay Quantities During Swings 341
5 Effect of Out-of-Step Conditions 343
5.1 Distance Relays 343
5.2 Directional Comparison Systems 344
5.3 Phase-Comparison or Pilot-Wire Systems 344
5.4 Underreaching Transfer-Trip Schemes 344
5.5 Circuit Breakers 344
5.6 Overcurrent Relays 344
5.7 Reclosing 344
6 Out-of-Step Relaying 345
6.1 Generator Out-of-Step Relaying 345
6.2 Transmission-Line Out-of-Step Relaying 346
7 Philosophies of Out-of-Step Relaying 346
7.1 Utility Practice 347
8 Types of Out-of-Step Schemes 347
8.1 Concentric Circle Scheme 347
8.2 Blinder Scheme 348
9 Relays for Out-of-Step Systems 348
9.1 Electromechanical Types 348
9.2 Solid-State Types 349
10 Selection of an Out-of-Step Relay System 351
15 Voltage Stability 353
L. Wang
1 Introduction 353
1.1 Small-Disturbance Instability 353
1.2 Large-Disturbance Instability 355
1.3 Voltage Instability Incidents 356
2 Voltage Instability Indices 357
2.1 Indices Based on Current Operating Condition 357
2.2 Indices Based on Stressed System Conditions 360
2.3 Summary 362
3 Voltage Instability Protection 362
3.1 Reactive Power Control 362
3.2 Load Tap Changer Blocking Schemes 362
3.3 Load Shedding 362
xiv Contents

16 Reclosing and Synchronizing 365

Revised by S. Ward
1 Introduction 365
2 Reclosing Precautions 365
3 Reclosing System Considerations 366
3.1 One-Shot vs. Multiple-Shot Reclosing Relays 366
3.2 Selective Reclosing 366
3.3 Deionizing Times for Three-Pole Reclosing 366
3.4 Synchronism Check 366
3.5 Live-Line/Dead-Bus, Live-Bus/Dead-Line Control 367
3.6 Instantaneous-Trip Lockout 367
3.7 Intermediate Lockout 367
3.8 Compatibility with Supervisory Control 367
3.9 Inhibit Control 368
3.10 Breaker Supervision Functions 368
3.11 Factors Governing Application of Reclosing 368
4 Considerations for Applications of Instantaneous Reclosing 368
4.1 Feeders with No-Fault-Power Back-Feed and Minimum Motor Load 369
4.2 Single Ties to Industrial Plants with Local Generation 369
4.3 Lines with Sources at Both Ends 369
5 Reclosing Relays and Their Operation 369
5.1 Review of Breaker Operation 369
5.2 Single-Shot Reclosing Relays 369
5.3 Multishot Reclosing Relays 371
6 Synchronism Check 377
6.1 Phasing Voltage Synchronism Check Characteristic 377
6.2 Angular Synchronism Check Characteristic 378
7 Dead-Line or Dead-Bus Reclosing 379
8 Automatic Synchronizing 379

17 Load-Shedding and Frequency Relaying 381

Revised by W. A. Elmore
1 Introduction 381
2 Rate of Frequency Decline 381
3 Load-Shedding 383
4 Frequency Relays 384
4.1 KF Induction-Cylinder Underfrequency Relay 384
4.2 Digital Frequency Relays 385
4.3 Microprocessor-Based Frequency Relay 385
5 Formulating a Load-Shedding Scheme 385
5.1 Maximum Anticipated Overload 385
5.2 Number of Load-Shedding Steps 386
5.3 Size of the Load Shed at Each Step 386
5.4 Frequency Settings 387
5.5 Time Delay 388
5.6 Location of the Frequency Relays 388
6 Special Considerations for Industrial Systems 389
Contents xv

7 Restoring Service 390

8 Other Frequency Relay Applications 391

Bibliography 395

Index 399
Introduction and General Philosophies
Revised by: W. A. ELMORE

1 INTRODUCTION Regulating relays. Are activated when an operat-

ing parameter deviates from predetermined
Relays are compact analog, digital, and numerical limits. Regulating relays function through sup-
devices that are connected throughout the power plementary equipment to restore the quantity to
system to detect intolerable or unwanted conditions the prescribed limits.
within an assigned area. They are, in effect, a form of Auxiliary relays. Operate in response to the open-
active insurance designed to maintain a high degree of ing or closing of the operating circuit to
service continuity and limit equipment damage. They supplement another relay or device. These
are ‘‘silent sentinels.’’ Although protective relays will include timers, contact-multiplier relays, sealing
be the main emphasis of this book, other types of units, isolating relays, lockout relays, closing
relays applied on a more limited basis or used as part relays, and trip relays.
of a total protective relay system will also be covered. Synchronizing (or synchronism check) relays. As-
sure that proper conditions exist for intercon-
necting two sections of a power system.

2 CLASSIFICATION OF RELAYS Many modern relays contain several varieties of these

functions. In addition to these functional categories,
Relays can be divided into six functional categories: relays may be classified by input, operating principle or
Protective relays. Detect defective lines, defective structure, and performance characteristic. The follow-
apparatus, or other dangerous or intolerable ing are some of the classifications and definitions
conditions. These relays generally trip one or described in ANSI/IEEE Standard C37.90 (see also
more circuit breaker, but may also be used to ANSI/IEEE C37.100 ‘‘Definitions for Power Switch-
sound an alarm. gear’’):
Monitoring relays. Verify conditions on the power
system or in the protection system. These relays Inputs
include fault detectors, alarm units, channel- Current
monitoring relays, synchronism verification, and Voltage
network phasing. Power system conditions that Power
do not involve opening circuit breakers during Pressure
faults can be monitored by verification relays. Frequency
Reclosing relays. Establish a closing sequence for a Temperature
circuit breaker following tripping by protective Flow
relays. Vibration

2 Chapter 1

Operating Principle or Structures 2.1.3 Numerical

Current balance
Numerical relays are those in which the measured ac
quantities are sequentially sampled and converted into
numeric data form. A microprocessor performs
mathematical and/or logical operations on the data
Solid state
to make trip decisions (e.g., MDAR, MSOC, DPU,
TPU, REL-356, REL-350, REL-512).
Performance Characteristics THEIR DESIGN
Distance Technically, most relays are small systems within
Directional overcurrent themselves. Throughout this book, however, the term
Inverse time system will be used to indicate a combination of relays
Definite time of the same or different types. Properly speaking, the
Undervoltage protective relaying system includes circuit breakers and
Overvoltage current transformers (ct’s) as well as relays. Relays,
Ground or phase ct’s, and circuit breakers must function together. There
High or low speed is little or no value in applying one without the other.
Pilot Protective relays or systems are not required to
Phase comparison function during normal power system operation, but
Directional comparison must be immediately available to handle intolerable
Current differential system conditions and avoid serious outages and
damage. Thus, the true operating life of these relays
A separate volume, Pilot Protective Relaying, covers
can be on the order of a few seconds, even though they
pilot systems (those relaying functions that involve a
are connected in a system for many years. In practice,
communications channel between stations.
the relays operate far more during testing and main-
tenance than in response to adverse service conditions.
2.1 Analog/Digital/Numerical In theory, a relay system should be able to respond
to an infinite number of abnormalities that can
Solid-state (and static) relays are further categorized possibly occur within the power system. In practice,
under one of the following designations. the relay engineer must arrive at a compromise based
on the four factors that influence any relay application:
2.1.1 Analog Economics. Initial, operating, and maintenance
Available measures of fault or troubles. Fault
Analog relays are those in which the measured magnitudes and location of current transformers
quantities are converted into lower voltage but similar and voltage transformers
signals, which are then combined or compared directly Operating practices. Conformity to standards and
to reference values in level detectors to produce the accepted practices, ensuring efficient system
desired output (e.g., SA-1 SOQ, SI-T, LCB, circuit operation
shield relays). Previous experience. History and anticipation of
the types of trouble likely to be encountered
2.1.2 Digital within the system
Digital relays are those in which the measured ac The third and fourth considerations are perhaps better
quantities are manipulated in analog form and expressed as the ‘‘personality of the system and the
subsequently converted into square-wave (binary) relay engineer.’’
voltages. Logic circuits or microprocessors compare Since it is simply not feasible to design a protective
the phase relationships of the square waves to make a relaying system capable of handling any potential
trip decision (e.g., SKD-T, REZ-1). problem, compromises must be made. In general, only
Introduction and General Philosophies 3

those problems that, according to past experience, are 3.1.1 Reliability

likely to occur receive primary consideration. Natu-
System reliability consists of two elements: depend-
rally, this makes relaying somewhat of an art. Different
ability and security. Dependability is the degree of
relay engineers will, using sound logic, design sig-
certainty of correct operation in response to system
nificantly different protective systems for essentially
trouble, whereas security is the degree of certainty that
the same power system. As a result, there is little
a relay will not operate incorrectly. Unfortunately,
standardization in protective relaying. Not only may
these two aspects of reliability tend to counter one
the type of relaying system vary, but so will the extent
another; increasing security tends to decrease depend-
of the protective coverage. Too much protection is
ability and vice versa. In general, however, modern
almost as bad as too little.
relaying systems are highly reliable and provide a
Nonetheless, protective relaying is a highly specia-
practical compromise between security and depend-
lized technology requiring an in-depth understanding
ability. The continuous supervision made possible by
of the power system as a whole. The relay engineer
numerical techniques affords improvement in both
must know not only the technology of the abnormal,
dependability and security. Protective relay systems
but have a basic understanding of all the system
must perform correctly under adverse system and
components and their operation in the system. Relay-
environmental conditions.
ing, then, is a ‘‘vertical’’ speciality requiring a
Dependability can be checked relatively easily in the
‘‘horizontal’’ viewpoint. This horizontal, or total
laboratory or during installation by simulated tests or
system, concept of relaying includes fault protection
a staged fault. Security, on the other hand, is much
and the performance of the protection system during
more difficult to check. A true test of system security
abnormal system operation such as severe overloads,
would have to measure response to an almost infinite
generation deficiency, out-of-step conditions, and so
variety of potential transients and counterfeit trouble
forth. Although these areas are vitally important to the
indications in the power system and its environment. A
relay engineer, his or her concern has not always been
secure system is usually the result of a good back-
fully appreciated or shared by colleagues. For this
ground in design, combined with extensive model
reason, close and continued communication between
power system or EMTP (electromagnetic transient
the planning, relay design, and operation departments
program) testing, and can only be confirmed in the
is essential. Frequent reviews of protective systems
power system itself and its environment.
should be mandatory, since power systems grow and
operating conditions change.
A complex relaying system may result from poor 3.1.2 Speed
system design or the economic need to use fewer circuit Relays that could anticipate a fault are utopian. But,
breakers. Considerable savings may be realized by even if available, they would doubtlessly raise the
using fewer circuit breakers and a more complex relay question of whether or not the fault or trouble really
system. Such systems usually involve design compro- required a trip-out. The development of faster relays
mises requiring careful evaluation if acceptable protec- must always be measured against the increased
tion is to be maintained. It should be recognized that probability of more unwanted or unexplained opera-
the exercise of the very best relaying application tions. Time is an excellent criterion for distinguishing
principles can never compensate for the absence of a between real and counterfeit trouble.
needed circuit breaker. Applied to a relay, high speed indicates that the
operating time usually does not exceed 50 ms (three
cycles on a 60-Hz base). The term instantaneous
3.1 Design Criteria indicates that no delay is purposely introduced in the
operation. In practice, the terms high speed and
The application logic of protective relays divides the instantaneous are frequently used interchangeably.
power system into several zones, each requiring its own
group of relays. In all cases, the four design criteria
3.1.3 Performance vs. Economics
listed below are common to any well-designed and
efficient protective system or system segment. Since it Relays having a clearly defined zone of protection
is impractical to satisfy fully all these design criteria provide better selectivity but generally cost more.
simultaneously, the necessary compromises must be High-speed relays offer greater service continuity by
evaluated on the basis of comparative risks. reducing fault damage and hazards to personnel, but
4 Chapter 1

also have a higher initial cost. The higher performance

and cost cannot always be justified. Consequently,
both low- and high-speed relays are used to protect
power systems. Both types have high reliability
records. Records on protective relay operations con-
sistently show 99.5% and better relay performance.

3.1.4 Simplicity
As in any other engineering discipline, simplicity in a
protective relay system is always the hallmark of
good design. The simplest relay system, however, is
not always the most economical. As previously
indicated, major economies may be possible with a
complex relay system that uses a minimum number
of circuit breakers. Other factors being equal,
simplicity of design improves system reliability—if
only because there are fewer elements that can Figure 1-1 A typical system and its zones of protection.
4. Transmission and distribution circuits
3.2 Factors Influencing Relay Performance 5. Motors

Relay performance is generally classed as (1) correct, A typical power system and its zones of protection are
(2) no conclusion, or (3) incorrect. Incorrect operation shown in Figure 1-1. The location of the current
may be either failure to trip or false tripping. The cause transformers supplying the relay or relay system
of incorrect operation may be (1) poor application, (2) defines the edge of the protective zone. The purpose
incorrect settings, (3) personnel error, or (4) equipment of the protective system is to provide the first line of
malfunction. Equipment that can cause an incorrect protection within the guidelines outlined above. Since
operation includes current transformers, voltage trans- failures do occur, however, some form of backup
formers, breakers, cable and wiring, relays, channels, protection is provided to trip out the adjacent breakers
or station batteries. or zones surrounding the trouble area.
Incorrect tripping of circuit breakers not associated Protection in each zone is overlapped to avoid the
with the trouble area is often as disastrous as a failure possibility of unprotected areas. This overlap is
to trip. Hence, special care must be taken in both accomplished by connecting the relays to current
application and installation to ensure against this. transformers, as shown in Figure 1-2a. It shows the
‘‘No conclusion’’ is the last resort when no evidence connection for ‘‘dead tank’’ breakers, and Figure 1-2b
is available for a correct or incorrect operation. Quite the ‘‘live tank’’ breakers commonly used with EHV
often this is a personnel involvement. circuits. Any trouble in the small area between the
current transformers will operate both zone A and B
relays and trip all breakers in the two zones. In
3.3 Zones of Protection Figure 1-2a, this small area represents the breaker, and
in Figure 1-2b the current transformer, which is
The general philosophy of relay applications is to generally not part of the breaker.
divide the power system into zones that can be
protected adequately with fault recognition and
removal producing disconnection of a minimum
amount of the system.
The power system is divided into protective zones
The first step in applying protective relays is to state
the protection problem accurately. Although develop-
1. Generators ing a clear, accurate statement of the problem can
2. Transformers often be the most difficult part, the time spent will pay
3. Buses dividends—particularly when assistance from others is
Introduction and General Philosophies 5

Transformer connections are particularly impor-

tant. For ground relaying, the location of all ground
‘‘sources’’ must also be known.

4.2 Existing System Protection and Procedures

The existing protective equipment and reasons for the

desired change(s) should be outlined. Deficiencies in
the present relaying system are a valuable guide to
improvements. New installations should be so speci-
fied. As new relay systems will often be required to
operate with or utilize parts of the existing relaying,
details on these existing systems are important.
Whenever possible, changes in system protection
should conform with existing operating procedures
and practices. Exceptions to standard procedures tend
to increase the risk of personnel error and may disrupt
the efficient operation of the system. Anticipated
system expansions can also greatly influence the choice
of protection.
Figure 1-2 The principle of overlapping protection around
a circuit breaker.
4.3 Degree of Protection Required
desired. Information on the following associated or
To determine the degree of protection required, the
supporting areas is necessary:
general type of protection being considered should be
System configuration outlined, together with the system conditions or
Existing system protection and any known deficien- operating procedures and practices that will influence
cies the final choice. These data will provide answers to the
Existing operating procedures and practices and following types of questions. Is pilot, high-, medium-,
possible future expansions or slow-speed relaying required? Is simultaneous
Degree of protection required tripping of all breakers of a transmission line required?
Fault study Is instantaneous reclosing needed? Are generator
Maximum load and current transformer locations neutral-to-ground faults to be detected?
and ratios
Voltage transformer locations, connections, and
4.4 Fault Study
Impedance of lines, transformers, and generators
An adequate fault study is necessary in almost all relay
4.1 System Configuration applications. Three-phase faults, line-to-ground faults,
and line-end faults should all be included in the study.
System configuration is represented by a single-line Line-end fault (fault on the line side of an open
diagram showing the area of the system involved in the breaker) data are important in cases where one breaker
protection application. This diagram should show in may operate before another. For ground-relaying, the
detail the location of the breakers; bus arrangements; fault study should include zero sequence currents and
taps on lines and their capacity; location and size of the voltages and negative sequence currents and voltages.
generation; location, size, and connections of the These quantities are easily obtained during the course
power transformers and capacitors; location and ratio of a fault study and are often extremely useful in
of ct’s and vt’s; and system frequency. solving a difficult relaying problem.
6 Chapter 1

4.5 Maximum Loads, Transformer Data, and


Maximum loads, current and voltage transformer

connections, ratios and locations, and dc voltage are
required for proper relay application. Maximum loads
should be consistent with the fault data and based on
the same system conditions. Line and transformer
impedances, transformer connections, and grounding
methods should also be known. Phase sequence should
be specified if three-line connection drawings are
Obviously, not all the above data are necessary in
every application. It is desirable, however, to review
the system with respect to the above points and,
wherever applicable, compile the necessary data.
In any event, no amount of data can ensure a
successful relay application unless the protection
problems are first defined. In fact, the application
Figure 1-3 A typical switchboard type relay. (The CR
problem is essentially solved when the available directional time overcurrent relay in the Flexitest case.)
measures for distinguishing between tolerable and
intolerable conditions can be identified and specified.
The important designations in the ac schematic for
the relay, such as that illustrated in Figure 1-5, are
Phase rotation
Connected to the power system through the current Tripping direction
and voltage transformers, protective relays are wired Current and voltage transformer polarities
into the control circuit to trip the proper circuit
breakers. In the following discussion, typical connec-
tions for relays mounted on conventional switchboards
and for rack-mounted solid-state relays will be used to
illustrate the standard application practices and

5.1 Switchboard Relays

Many relays are supplied in a rectangular case that is

permanently mounted on a switchboard located in the
substation control house. The relay chassis, in some
implementations, slides into the case and can be
conveniently removed for testing and maintenance.
The case is usually mounted flush and permanently
wired to the input and control circuits. In the Flexitest
case, the electrical connections are made through
small, front-accessible, knife-blade switches. A typical
switchboard relay is shown in Figure 1-3; its corre-
sponding internal schematic is shown in Figure 1-4.
While the example shown is an electromechanical Figure 1-4 Typical internal schematic for a switchboard-
relay, many solid-state relays are in the Flexitest case mounted relay. (The circuit shown is for the CR directional
for switchboard mounting. time overcurrent relay of Figure 1-3.)
Introduction and General Philosophies 7

Figure 1-6 Typical dc schematic for a switchboard-

mounted relay. (The connections are for three phase type
CR and one CRC ground directional time overcurrent relays
of Figure 1-3 applied to trip a circuit breaker.)
Figure 1-5 Typical ac schematic for a switchboard-
mounted relay. (The connections are for the CR phase and the breaker. Line voltage cannot be used directly since,
CRC ground directional time overcurrent relay of Figure 1-3.) of course, it may be quite low during fault conditions.

Relay polarity and terminal numbers 5.2 Rack-Mounted Relays

Phasor diagram
Solid-state and microprocessor relays are usually rack-
All these designations are required for a directional mounted (Fig. 1-8). Since these relays involve more
relay. In other applications, some may not apply. In complex and sophisticated circuitry, different levels of
accordance with convention, all relay contacts are information are required to understand their opera-
shown in the position they assume when the relay is tion. A block diagram provides understanding of the
deenergized. basic process. Figure 1-9 is a block diagram for the
A typical control circuit is shown in Figure 1-6. MDAR microprocessor relay. Detailed logic diagrams
Three phase relays and one ground relay are shown plus ac and dc schematics are also required for a
protecting this circuit. Any one could trip the complete view of the action to be expected from these
associated circuit breaker to isolate the trouble or relays.
fault area. A station battery, either 125 Vdc or 250 Vdc,
is commonly used for tripping. Lower-voltage batteries
are not recommended for tripping service when long
trip leads are involved.
In small stations where a battery cannot be justified,
tripping energy is obtained from a capacitor trip
device. This device is simply a capacitor charged,
through a rectifier, by the ac line voltage. An example
of this arrangement is presented in Figure 1-7. When
the relay contacts close, the discharge of the energy in
the capacitor through the trip coil is sufficient to trip Figure 1-7 Typical capacitor trip device schematic.
8 Chapter 1


Complete tripping and closing circuits for circuit

breakers are complex. A typical circuit diagram is
shown in Figure 1-10. In this diagram, the protective
relay circuits, such as that shown in Figure 1-6, are
abbreviated to a single contact marked ‘‘prot relays.’’
While the trip circuits must be energized from a source
available during a fault (usually the station battery),
the closing circuits may be operated on ac. Such
breakers have control circuits similar to those shown in
Figure 1-10, except that the 52X, 52Y, and 52CC
circuits are arranged for ac operation.
The scheme shown includes red light supervision of
Figure 1-8 A typical rack type relay. (The SBFU static the trip coil, 52X/52Y antipump control, and low-
circuit breaker failure relay.) pressure and latch checks that most breakers contain in
some form.

Figure 1-9 Block diagram of MDAR relay.

Introduction and General Philosophies 9

Table 1-1 Comparison of Symbols

U.S. European
Element practice practice

Normally open contact

Normally closed contact

Form C



Current transformer


Figure 1-10 A typical control circuit schematic for a circuit Phase designations (typical) A,B,C RST
breaker showing the tripping and closing circuits. (preferred)
1, 2, 3
Component designations 1, 2, 0 1, 2, 0
(positive, negative, zero)
Voltage V U
Various symbols are used throughout the world to
represent elements of the power system. Table 1-1
compiles a few of the differences.
Technical Tools of the Relay Engineer: Phasors, Polarity, and
Symmetrical Components
Revised by: W. A. ELMORE

1 INTRODUCTION There are several systems and many variations of

phasor notation in use. The system outlined below is
In addition to a general knowledge of electrical power standard with most relay manufacturers.
systems, the relay engineer must have a good working
understanding of phasors, polarity, and symmetrical
components, including voltage and current phasors
during fault conditions. These technical tools are used
2.1 Circuit Diagram Notation for Current and
for application, analysis, checking, and testing of
protective relays and relay systems.
The reference direction for the current or flux can be
indicated by (1) an identified directional arrow in the
circuit diagram, as shown in Figure 2-1, or (2) the
2 PHASORS double subscript method, such as Iab, defined as the
current flowing from terminal a to terminal b, as in
A phasor is a complex number used to represent Figure 2-2.
electrical quantities. Originally called vectors, the In all cases, the directional arrow or double
quantities were renamed to avoid confusion with space subscript indicates the actual or assumed direction of
vectors. A phasor rotates with the passage of time and current (or flux) flow through the circuit during the
represents a sinusoidal quantity. A vector is stationary positive half-cycle of the ac wave.
in space.
In relaying, phasors and phasor diagrams are used
both to aid in applying and connecting relays and for
the analysis of relay operation after faults.
Phasor diagrams must be accompanied by a circuit
diagram. If not, then such a circuit diagram must be
obvious or assumed in order to interpret the phasor
diagram. The phasor diagram shows only the magni-
tude and relative phase angle of the currents and
voltages, whereas the circuit diagram illustrates only
the location, direction, and polarity of the currents and
voltages. These distinctions are important. Confusion
generally results when the circuit diagram is omitted or Figure 2-1 Reference circuit diagram illustrating single
the two diagrams are combined. subscript notation.

12 Chapter 2

2.3 Phasor Notation

Figure 2.3a demonstrates the relationship between a

phasor and the sinusoid it represents. At a chosen time
(in this instance at the time at which the phasor has
advanced to 308), the instantaneous value of the
sinusoid is the projection on the vertical of the point
of the phasor.
Phasors must be referred to some reference frame.
Figure 2-2 Reference circuit diagram illustrating double The most common reference frame consists of the axis
subscript notation. (Current arrows not required but are of real quantities x and the axis of imaginary quantities
usually shown in practice.)
y, as shown in Figure 2-3b. The axes are fixed in the
plane, and the phasors rotate, since they are sinusoidal
quantities. (The convention for positive rotation is
counterclockwise.) The phasor diagram therefore
represents the various phasors at any given common
instant of time.
Theoretically, the length of a phasor is proportional
2.2 Circuit Diagram Notation for Voltage to its maximum value, with its projections on the real
and imaginary axes representing its real and imaginary
The relative polarity of an ac voltage may be shown in components at that instant. By arbitrary convention,
the circuit diagram by (1) a þ mark at one end of the however, the phasor diagram is constructed on the
locating arrow (Fig. 2-1) or (2) the double subscript
notation (Fig. 2-2). In either case, the meaning of the
notation must be clearly understood. Failure to
properly define notation is the basis for much
confusion among students and engineers.
The notation used in this text is defined as follows:

The letter ‘‘V’’ is used to designate voltages. For

simplicity, only voltage drops are used. In this
sense, a generator rise is considered a negative
drop. Some users assign the letter ‘‘E’’ to
generated voltage. In much of the world, ‘‘U’’
is used for voltage.
If locating arrows are used for voltage in the circuit Figure 2-3a Phasor generation of sinusoid.
diagram with a single subscript notation, the þ
mark at one end indicates the terminal of actual
or assumed positive potential relative to the other
in the half-cycle.
If double subscript notation is used, the order of the
subscripts indicates the actual or assumed direc-
tion of the voltage drop when the voltage is in the
positive half-cycle.

Thus, the voltage between terminals a and b may be

written as either Vab or Eab . Voltage Vab or Eab is
positive if terminal a is at a higher potential than
terminal b when the ac wave is in the positive half-
cycle. During the negative half-cycle of the ac wave,
Vab or Eab is negative, and the actual drop for that
half-cycle is from terminal b to terminal a. Figure 2-3b Reference axis and nomenclature for phasors.
Phasors, Polarity, and Symmetrical Components 13

basis of rms values, which are used much more 2.3.2 Division Law
frequently than maximum values. The phasor diagram
The division law is the inverse of multiplication:
indicates angular relationships under the chosen
conditions, normal or abnormal. E jEjejy1 jEj
For reference and review, the various forms, for ¼ ¼ ffðy1  y2 Þ ð2-8Þ
I jIjejy2 jIj
representation of point P in Figure 2-3b are as follows:

Rectan- Expo-
gular Complex nential Polar Phasor 2.3.3 Powers of Complex Numbers
form form form form form
a þ jb ¼ jcjðcos y þ j sin yÞ ¼ jcjejy ¼ jcjffy ¼ c ð2-1Þ The product of a phasor times its conjugate is
a  jb ¼ jcjðcos y  j sin yÞ ¼ jcje 
¼ jcjffy ¼ ^c ð2-2Þ ðjIjejy ÞZ ¼ jIjZ ejZy ð2-9Þ
2 j2y
where Thus, I2 equals jIj e :
a ¼ real value qffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffi jy 
jIjejy ¼ Z jIj e Z ð2-10Þ
b ¼ imaginary value
jcj ¼ modulus or absolute value ðmagnitudeÞ
The product of a phasor times its conjugate is
y ¼ argument or amplitude ðrelative positionÞ
If c is a phasor, then ^c is its conjugate. Thus, if I^I ¼ jIjejy 6jIjejy

c ¼ a þ jb ¼ jIj2 ejðyyÞ
¼ jIj2 ð2-11Þ
^c ¼ a  jb Other reference axes used frequently are shown in
Figure 2-4. Their application will be covered in later
Some references use c* to represent conjugate. chapters.
The absolute value of the phasor is jcj:
jcj ¼ a2 þ b2 ð2-3Þ 2.4 Phasor Diagram Notation
By adding Eqs. (2-1) and (2-2), we obtain
In Figure 2-5, the phasors all originate from a common
1 origin. This method is preferred. In an alternative
a ¼ ðc þ ^cÞ ð2-4Þ
2 method, shown in Figure 2-6, the voltage phasors are
Substracting Eqs. (2-1) and (2-2) yields moved away from a common origin to illustrate the
phasor addition of voltages in series (closed system).
1 Although this diagram notation can be useful, it is not
jb ¼ ðc  ^cÞ ð2-5Þ
In addition to the use of a single term such as c for a
phasor, c_; c, and *
c have also been used.

2.3.1 Multiplication Law

The absolute value of a phasor product is the product
of the absolute values of its components, and the
argument is the sum of the component arguments:
EI ¼ jEj6jIjffðy1 þ y2 Þ ð2-6Þ

E^I ¼ jEjejy1 6jIjejy2

Figure 2-4 Other reference axes for phasors used in relaying
¼ jEj6jIjffðy1  y2 Þ ð2-7Þ and power systems.
14 Chapter 2

Figure 2-5 Open-type phasor diagram for the basic

elements (resistor, reactor, and capacitor) connected in series.

generally recommended since it often promotes confu-

sion by combining the circuit and phasor diagrams.
Notation for three-phase systems varies consider-
ably in the United States; the phases are labeled a, b, c
or A, B, C or 1, 2, 3. In other countries, the
corresponding phase designation of r, s, t is frequently
The letter designations are preferred and used here to
Figure 2-7 Designation of the voltages and currents in a
avoid possible confusion with symmetrical components
three-phase power system.
notation. A typical three-phase system, with its separate
circuit and phasor diagrams, is shown in Figure 2-7.
The alternative closed-system phasor diagram is shown
Similarly, Vbc ¼ Vbn  Vcn , and Vca ¼ Vcn  Van . The
in Figure 2-8. With this type of diagram, one tends to
associated phasors are shown in Figure 2-7.
label the three corners of the triangle a, b, and c—
Neutral (n) and ground (g) are often incorrectly
thereby combining the circuit and phasor diagrams.
interchanged. They are not the same. The voltage from
The resulting confusion is apparent when one notes
n to g is zero when no zero sequence voltage exists.
that, with a at the top corner and b at the lower right
With zero sequence current flowing, there will be a
corner, the voltage drop from a to b would indicate the
voltage between neutral and ground, Vng ¼ Vo .
opposite arrow from that shown on Vab .
However, when it is considered that, always,
Vab ¼ Van þ Vnb , it is evident that Vab ¼ Van  Vbn .

Figure 2-6 Alternative closed-type phasor diagram for the Figure 2-8 Alternative closed system phasor diagram for
basic circuit of Figure 2-5. the three-phase power system of Figure 2-7.
Phasors, Polarity, and Symmetrical Components 15

Ground impedance (Rg or RL ) resulting in a rise in

station ground potential can be an important factor in
relaying. This will be considered in later chapters.
According to ANSI/IEEE Standard 100, ‘‘the neutral
point of a system is that point which has the same
potential as the point of junction of a group of equal
nonreactive resistances if connected at their free ends to
the appropriate main terminals or lines of the system.’’
Figure 2-10 Polarity and circuit diagram for conventional
representation of current and linear coupler transformer.
2.5 Phase Rotation vs. Phasor Rotation

Phase rotation, or preferably phase sequence, is the

order in which successive phase phasors reach their
positive maximum values. Phasor rotation is, by
international convention, counterclockwise in direc- & indicate
The polarity marks X or ——
tion. Phase sequence is the order in which the phasors
pass a fixed point. The current flowing out at the polarity-marked
All standard relay diagrams are for phase rotation terminal on the secondary side is essentially in
a, b, c. It is not uncommon for power systems to have phase with the current flowing in at the polarity-
one or more voltage levels with a, c, b rotation; then marked terminal on the primary side.
specific diagrams must be made accordingly. The The voltage drop from the polarity-marked to the
connection can be changed from one rotation to the non-polarity-marked terminal on the primary
other by completely interchanging all b and c connec- side is essentially in phase with the voltage drop
tions. from the polarity-marked to the non-polarity-
marked terminals on the secondary side.

3 POLARITY IN RELAY CIRCUITS The expression ‘‘essentially in phase’’ allows for the
small phase-angle error.
3.1 Polarity of Transformers

The polarity indications shown in Figures 2-9 and 2-10

apply for both current and voltage transformers, or
any type of transformer with either subtractive or 3.2 Polarity of Protective Relays
additive polarity.
Polarity is always associated with directional-type
relay units, such as those indicating the direction of
power flow. Other protective relays, such as distance
types, may also have polarity markings associated with
their operation. Relay polarity is indicated on the
schematic or wiring diagrams by a small þ mark above
or near the terminal symbol or relay winding. Two
such marks are necessary; a mark on one winding
alone has no meaning.
Typical polarity markings for a directional unit are
shown in Figure 2-11. In this example, the markings
indicate that the relay will operate when the voltage
drop from polarity to nonpolarity in the voltage coil is
in phase with the current flow from polarity to
nonpolarity in the current coil. This applies irrespec-
tive of the maximum sensitivity angle of the relay. Of
course, the levels must be above the relay pickup
Figure 2-9 Polarity and circuit diagram for transformers. quantities for the relay to operate.
16 Chapter 2

zero torque line is a zone of no operation and not a

thin line through the origin, as commonly drawn.

3.3.2 Ground Directional Unit

As shown in Figure 2-13, the ground directional unit
usually has a characteristic of maximum torque when I
flowing from polarity to nonpolarity lags V drop from
Figure 2-11 Polarity markings for protective relays.
polarity to nonpolarity by 608. Although this char-
acteristic may be inherent in the unit’s design, an
auxiliary phase shifter is generally required in analog
3.3 Characteristics of Directional Relays
Directional units are often used to supervise the action
of fault responsive devices such as overcurrent units.
The primary function of the directional units is to limit
3.3.3 Watt-Type Directional Unit
relay operation to a specified direction. These highly
sensitive units operate on load in the tripping direction. The characteristic of the watt-type unit is as shown in
Directional units can conveniently serve to illustrate Figure 2-14. It has maximum sensitivity when relay
the practical application of phasors and polarity. In current and voltage are in phase.
addition to polarity, these units have a phase-angle
characteristic that must be understood if they are to be
properly connected to the power system. The char-
acteristics discussed below are among the most

3.3.1 Cylinder-Type Directional Unit

As shown in Figure 2-12, the cylinder-type unit has
maximum torque when I, flowing in the relay winding
from polarity to nonpolarity, leads V drop from
polarity to nonpolarity by 308. The relay minimum
pickup values are normally specified at this maximum
torque angle. As current Ipq lags or leads this
maximum torque position, more current is required Figure 2-13 Phase-angle characteristics of a ground direc-
(at a constant voltage) to produce the same torque. tional relay unit.
Theoretically, at 1208 lead or 608 lag, no torque results
from any current magnitude. In practice, however, this

Figure 2-12 Phase-angle characteristics of the cylinder-type Figure 2-14 Phase-angle characteristics of a watt-type
directional relay unit. directional relay unit.
Phasors, Polarity, and Symmetrical Components 17

3.4 Connections of Directional Units to Three- quantities to apply to the directional units. For all
Phase Power Systems faults in the operating zone of the relay, the fault
current and voltage should produce an operating
The relay unit’s individual characteristic, as discussed condition as close to maximum sensitivity as possible.
so far, is the characteristic that would be measured on Fault current generally lags its unity power factor
a single-phase test. Faults on three-phase power position by 20 to 858, depending on the system voltage
systems can, however, produce various relations and characteristics.
between the voltages and currents. To ensure correct Four types of directional element connections (Fig.
relay operation, it is necessary to select the proper 2-15) have been used for many years. The proper

Figure 2-15 Directional element connections.

18 Chapter 2

system quantities are selected to yield the best electrical, mechanical, and thermal failure or any
operation, considering the phase-angle characteristic combination of these.
of the directional unit. A study of these connections
reveals that none is perfect. All will provide incorrect
4.1 Fault Types and Causes
operation under some fault conditions. These condi-
tions are, moreover, different for each connection.
To ensure adequate protection, the conditions existing
Fortunately, the probability of such fault conditions
on a system during faults must be clearly understood.
occurring in most power systems is usually very low.
These abnormal conditions provide the discriminating
For phase directional measurements, the standard
means for relay operation. The major types and causes
908 connection is the one best suited to most power
of failure are listed in Table 2-1.
systems. Here, the system quantities applied to the relay
Relays must operate for several types of faults:
are 908 apart at unity power factor, balanced current.
With this connection, maximum sensitivity can occur at Three-phase (a-b-c, a-b-c-g)
various angles, depending on relay design, as in Phase-to-phase (a-b, b-c, c-a)
connection 4. The 908 connection is one standard for Two-phase-to-ground (a-b-g, b-c-g, c-a-g)
phase relays. The 908 angle is that between the unity Phase-to-ground (a-g, b-g, c-g)
power factor current and the voltage applied to the relay.
Unless preceded by or caused by a fault, open
Some experts use a dual numbered system to describe
circuits on power systems occur infrequently. Conse-
the relationship of the system quantities and to identify
quently, very few relay systems are designed specifi-
the nature of the relaying unit itself. For example, the
cally to provide open-circuit protection. One exception
90–608 connection is one in which the unity power-
is in the lower-voltage areas, where a fuse can be open.
factor current applied to the relay and flowing in the
Another is in EHV, where breakers are equipped with
relay trip direction leads the voltage applied to the relay
independent pole mechanisms.
by 908. The nature of the relay referred to is such that
Simultaneous faults in two parts of the system are
the maximum sensitivity occurs when the system
generally impossible to relay properly under all
current lags its unity power phase position by 608.
conditions. If both simultaneous faults are in the
The relay has its maximum sensitivity in this case when
relays’ operating zone, at least one set of relays is likely
the current applied to it (into the polarity marker and
to operate, with the subsequent sequential operation
out nonpolarity) leads the voltage applied to it (voltage
of other relays seeing the faults. When faults appear
drop polarity to nonpolarity) by 308. Since this is
both internal and external simultaneously, some relays
somewhat confusing, it is recommended that the system
have difficulty determining whether to trip or not.
quantities that are applied to the relay be defined
Fortunately, simultaneous faults do not happen very
independent of the characteristics of the relay, and that
the characteristics of the relay be described independent
of the system quantities with which it is used. Table 2-1 Major Types and Causes of Failures
Figure 2-16 is a composite circuit diagram illustrat- Type Cause
ing the ‘‘phase-a’’ connections for these four connec-
tions that have been used over the years, together with Insulation Design defects or errors
the connection for a ground directional relay. The Improper manufacturing
Improper installation
phasor diagrams are shown in Figure 2-15a for the
Aging insulation
phase relays and Figure 2-17 for a commonly used Contamination
ground relay. Electrical Lightning surges
Switching surges
Dynamic overvoltages
4 FAULTS ON POWER SYSTEMS Thermal Coolant failure
A fault-proof power system is neither practical nor Overvoltage
economical. Modern power systems, constructed with Ambient temperatures
as high an insulation level as practical, have sufficient Mechanical Overcurrent forces
flexibility so that one or more components may be out Earthquake
Foreign object impact
of service with minimum interruption of service. In
Snow or ice
addition to insulation failure, faults can result from
Phasors, Polarity, and Symmetrical Components 19

Figure 2-16 Directional unit connections (phase ‘‘a’’ only) for four types of connections plus the ground directional relay
20 Chapter 2

Figure 2-17 Phasor diagram for the ground directional

relay connection shown in Figure 2-16. (Phase ‘‘a’’-to-ground
fault is assumed on a solidly grounded system.)

Figure 2-18 Voltage plot for a solid phase ‘‘a’’-to-ground

often and are not a significant cause of incorrect fault on an ungrounded system.

In a symmetrical system, where the three capaci-

4.2 Characteristics of Faults tances to ground are equal, g equals n. If phase a is
4.2.1 Fault Angles grounded, the triangle shifts as shown in Figure 2-18.
Consequently, Vbg and Vcg become approximately 3
The power factor, or angle of the fault current, is times their normal value. In contrast, a ground on one
determined for phase faults by the nature of the source phase of a solidly grounded radial system will result in
and connected circuits up to the fault location and, for a large phase and ground fault current, but little or no
ground faults, by the type of system grounding as well. increase in voltage on the unfaulted phases (Fig. 2-19).
The current will have an angle of 80 to 858 lag for a
phase fault at or near generator units. The angle will be
less out in the system, where lines are involved. 4.2.3 Fault Resistance
Typical open-wire transmission line angles are as Unless the fault is solid, an arc whose resistance varies
follows: with the arc length and magnitude of the fault current
7.2 to 23 kV: 20 to 458 lag is usually drawn through air. Several studies indicate
23 to 69 kV: 45 to 758 lag that for currents in excess of 100 A the voltage across
69 to 230 kV: 60 to 808 lag the arc is nearly constant at an average of approxi-
230 kV and up: 75 to 858 lag mately 440 V/ft.
Arc resistance is seldom an important factor in
At these voltage levels, the currents for phase faults phase faults except at low system voltages. The arc
will have the angles shown where the line impedance does not elongate sufficiently for the phase spacings
predominates. If the transformer and generator impe- involved to decrease the current flow materially. In
dances predominate, the fault angles will be higher. addition, the arc resistance is at right angles to the
Systems with cables will have lower angles if the cable reactance and, hence, may not greatly increase the total
impedance is a large part of the total impedance to the impedance that limits the fault current.

4.2.2 System Grounding

System grounding significantly affects both the magni-
tude and angle of ground faults. There are three classes
of grounding: ungrounded (isolated neutral), impe-
dance-grounded (resistance or reactance), and effec-
tively grounded (neutral solidly grounded). An
ungrounded system is connected to ground through
the natural shunt capacitance, as illustrated in Figure
2-18 (see also Chap. 7). In addition to load, small Figure 2-19 Voltage plot for a solid phase ‘‘a’’-to-ground
(usually negligible) charging currents flow normally. fault on a solidly grounded system.
Phasors, Polarity, and Symmetrical Components 21

For ground faults, arc resistance may be an other locations will be between these extremes,
important factor because of the longer arcs that can depending on the point of measurement.
occur. Also, the relatively high tower footing resistance
may appreciably limit the fault current.
Arc resistance is discussed in more detail in Chapter 5 SYMMETRICAL COMPONENTS
Relay application requires a knowledge of system
4.2.4 Distortion of Phases During Faults conditions during faults, including the magnitude,
direction, and distribution of fault currents, and often
The phasor diagrams in Figure 2-20 illustrate the effect
the voltages at the relay locations for various operating
of faults on the system voltages and currents. The conditions. Among the operating conditions to be
diagrams shown are for effectively grounded systems. considered are maximum and minimum generation,
In all cases, the dotted or uncollapsed voltage triangle
selected lines out, line-end faults with the adjacent
exists in the source (the generator) and the maximum
breaker open, and so forth. With this information, the
collapse occurs at the fault location. The voltage at
relay engineer can select the proper relays and settings
to protect all parts of the power system in a minimum
amount of time. Three-phase fault data are used for
the application and setting of phase relays and single-
phase-to-ground fault data for ground relays.
The method of symmetrical components is the
foundation for obtaining and understanding fault
data on three-phase power systems. Formulated by
Dr. C. L. Fortescue in a classic AIEE paper in 1918,
the symmetrical components method was given its first
practical application to system fault analysis by C. F.
Wagner and R. D. Evans in the late 1920s and early
1930s. W. A. Lewis and E. L. Harder added
measurably to its development in the 1930s.
Today, fault studies are commonly made with the
digital computer and can be updated rapidly in
response to system changes. Manual calculations are
practical only for simple cases.
A knowledge of symmetrical components is impor-
tant in both making a study and understanding the
data obtained. It is also extremely valuable in
analyzing faults and relay operations. A number of
protective relays are based on symmetrical compo-
nents, so the method must be understood in order to
apply these relays successfully.
In short, the method of symmetrical components is
one of the relay engineer’s most powerful technical
tools. Although the method and mathematics are quite
simple, the practical value lies in the ability to think
and visualize in symmetrical components. This skill
requires practice and experience.

5.1 Basic Concepts

The method of symmetrical components consists

Figure 2-20 Phasor diagrams for the various types of faults of reducing any unbalanced three-phase system of
occurring on a typical power system. phasors into three balanced or symmetrical systems:
22 Chapter 2

the positive, negative, and zero sequence components. change in magnitude:

This reduction can be performed in terms of current,
voltage, impedance, and so on. a ¼ 1ff120
The positive sequence components consist of three ¼ 0:5 þ j0:866 ð2-13Þ
phasors equal in magnitude and 1208 out of phase (Fig. 
a ¼ 1ff240
2-21a). The negative sequence components are three
¼ 0:5  j0:866 ð2-14Þ
phasors equal in magnitude, displaced 1208 with a

phase sequence opposite to that of the positive a ¼ 1ff360

sequence (Fig. 2-21b). The zero sequence components ¼ 1:0 þ j0 ð2-15Þ

consist of three phasors equal in magnitude and in
phase (Fig. 2-21c). Note all phasors rotate in a From these equations, useful combinations can be
counterclockwise direction. derived
In the following discussion, the subscript 1 will
identify the positive sequence component, the subscript 1 þ a þ a2 ¼ 0
2 the negative sequence component, and the subscript 0 1  a2 ¼ 3ff30 ð2-16Þ
the zero sequence component. For example, Va1 is the
positive sequence component of phase-a voltage, Vb2 or
the negative sequence component of phase-b voltage, pffiffiffi
a2  1 ¼3ff210
and Vc0 the zero sequence component of phase-c pffiffiffi
voltage. All components are phasor quantities, rotating a  1 ¼ 3ff150 ð2-17Þ
Since the three phasors in any set are always equal or
in magnitude, the three sets can be expressed in terms pffiffiffi
1a¼ 3ff  30
of one phasor. For convenience, the phase-a phasor is p ffiffi

used as a reference. Thus, a2  a ¼ 3ff270 ð2-18Þ

Positive Negative Zero pffiffiffi 
sequence sequence sequence a  a2 ¼ 3ff90 ð2-19Þ
Va1 ¼ Va1 Va2 ¼ Va2 Va0 ¼ Va0 ð2-12Þ Any three-phase system of phasors will always be
Vb2 ¼ aVa2 Vb0 ¼ Va0 the sum of the three components:
Vb1 ¼ a2 Va1
Vc0 ¼ Va0 Va ¼ Va1 þ Va2 þ Va0 ð2-20Þ
Vc1 ¼ aVa1 Vc2 ¼ a2 Va2
Vb ¼ Vb1 þ Vb2 þ Vb0
¼ a2 Va1 þ aVa2 þ Va0 ð2-21Þ
The coefficients a and a2 are operators that, when
multiplied with a phasor, result in a counterclockwise Vc ¼ Vc1 þ Vc2 þ Vc0
angular shift of 120 and 2408, respectively, with no ¼ aVa1 þ a2 Va2 þ Va0 ð2-22Þ

Since phase a has been chosen as a reference, the

subscripts are often dropped for convenience. Thus,
Va ¼ V1 þ V2 þ V0
Ia ¼ I1 þ I2 þ I0
Vb ¼ a2 V1 þ aV2 þ V0


Ib ¼ a2 I1 þ aI2 þ I0
Figure 2-21 Sequence components of voltages. Vc ¼ aV1 þ a2 V2 þ V0
Phasors, Polarity, and Symmetrical Components 23

Ic ¼ aI1 þ a2 I2 þ I0 ð2-25Þ
Quantities V1 ; V2 ; V0 ; I1 ; I2 , and I0 , can always be
assumed to be the phase-a components. Note that the
b and c components always exist, as indicated by Eq.
(2-12). Note that dropping the phase subscripts should
be done with great care. Where any possibility of
misunderstanding can occur, the additional effort of
using the double subscripts will be rewarded. Figure 2-22 Power system neutral.
Equations (2-20) to (2-22) can be solved to yield the
sequence components for a general set of three-phase
phasors: From Eq. (2-28),
1 1
Va1 ¼ ðVag þ aVbg þ a2 Vcg Þ V0 ¼ ðVag þ Vbg þ Vcg Þ
3 3
and Substituting Eq. (2-29), we obtain
1 1
Ia1 ¼ ðIa þ aIb þ a2 Ic Þ V0 ¼ ðVan þ Vng þ Vbn þ Vng þ Vcn þ Vng Þ
3 3
1 Since Van þ Vbn þ Vcn ¼ 0,
Va2 ¼ ðVag þ a2 Vbg þ aVcg Þ ð2-26Þ
and V0 ¼ ð3Vng Þ
1 V0 ¼ Vng
Ia2 ¼ ðIa þ a2 Ib þ aIc Þ
Neutral and ground are distinctly independent and
1 differ in voltage by V0 .
Va0 ¼ ðVag þ Vbg þ Vcg Þ ð2-27Þ
3 Grounding and its influence on relaying are
and discussed in Chapters 7 and 12.

I0 ¼ ðIa þ Ib þ Ic Þ ð2-28Þ
A sequence component cannot exist in only one 5.3 Sequences in a Three-Phase Power System
phase. If any sequence component exists by measure-
ment or calculation in one phase, it exists in all three Several important assumptions are made to greatly
phases, as shown in Eq. (2-12) and Figure 2-21. simplify the use of symmetrical components in
practical circumstances. Interconnections of the three
sequence networks allow any series or shunt disconti-
nuity to be investigated. For the rest of the power-
5.2 System Neutral
system network, it is assumed that the impedances in
the individual phases are equal and the generator
Figure 2-22 describes the definition of power-system
phase voltages are equal in magnitude and displaced
neutral and contrasts it with ground. Neutral is
1208 from one another.
established by connecting together the terminals of
Based on this premise, in the symmetrical part of the
three equal resistances as shown with each of the other system, positive sequence current flow produces only
resistor terminals connected to one of the phases. We
positive sequence voltage drops, negative sequence
can thus write
current flow produces only negative sequence voltage
Vag ¼ Van þ Vng drops, and zero sequence current flow produces only
zero sequence voltage drops. For an unsymmetrical
Vbg ¼ Vbn þ Vng
system, interaction occurs between components. For a
Vcg ¼ Vcn þ Vng ð2-29Þ particular series or shunt discontinuity being repre-
24 Chapter 2

sented, the interconnection of the networks produces magnitude ½3=ðx1 þ x2 þ x0 Þ greater than the three-
the required interaction. phase fault current magnitude ð1=x1 Þ. Since the
Any circuit that is not continuously transposed will machine is braced for only three-phase fault current
have impedances in the individual phases that differ. magnitude, it is seldom possible or desirable to ground
This fact is generally ignored in making calculations the neutral solidly.
because of the immense simplification that results. The armature winding resistance is small enough to
From a practical viewpoint, ignoring this effect, in be neglected in calculating short-circuit currents. This
general, has no appreciable influence. resistance is, however, important in determining the dc
time constant of an asymmetrical short-circuit current.
Typical reactance values for synchronous machin-
5.4 Sequence Impedances
ery are available from the manufacturer or handbooks.
However, actual design values should be used when
Quantities Z1 , Z2 , and Z0 are the system impedances to
the flow of positive, negative, and zero sequence
currents, respectively. Except in the area of a fault or
general unbalance, each sequence impedance is con- 5.4.2 Transformers
sidered to be the same in all three phases of the
symmetrical system. A brief review of these quantities The positive and negative sequence reactances of all
is given below for synchronous machinery, transfor- transformers are identical. Values are available from
mers, and transmission lines. the nameplate. The zero sequence reactance is either
equal to the other two sequence reactances or infinite
5.4.1 Synchronous Machinery except for the three-phase, core-type transformers. In
effect, the magnetic circuit design of the latter units
Three different positive sequence reactance values are gives them the effect of an additional closed delta
specified. X00d indicates the subtransient reactance, X0d winding. The resistance of the windings is very small
the transient reactance, and Xd the synchronous and neglected in short-circuit calculations.
reactance. These direct-axis values are necessary for The sequence circuits for a number of transformer
calculating the short-circuit current value at different banks are shown in Figure 2-23. The impedances
times after the short circuit occurs. Since the sub- indicated are the equivalent leakage impedances
transient reactance values give the highest initial between the windings involved. For two-winding
current value, they are generally used in system transformers, the total leakage impedance ZLH is
short-circuit calculations for high-speed relay applica- measured from the L winding, with the H winding
tion. The transient reactance value is used for stability short-circuited. ZHL is measured from the H winding
consideration and slow-speed relay application. with the L winding shorted. Except for a 1:1 transfor-
The unsaturated synchronous reactance is used for mer ratio, the impedances have different values in
sustained fault-current calculation since the voltage is ohms. On a per unit basis, however, ZLH equals ZHL .
reduced below saturation during faults near the unit. For three-winding and autotransformer banks,
Since this generator reactance is invariably greater there are three leakage impedances:
than 100%, the sustained fault current will be less than
the machine rated load current unless the voltage
regulator boosts the field substantially. Winding
The negative sequence reactance of a turbine measured Shorted Open
generator is generally equal to the subtransient X00d Impedance from winding winding
reactance. X2 for a salient-pole generator is much
higher. The flow of negative sequence current of ZHM(ZHL) H M(L) L(T)
opposite phase rotation through the machine stator
winding produces a double frequency component in
the rotor. As a result, the average of the subtransient
direct-axis reactance and the subtransient quadrature- Both winding conventions shown above are in
axis reactance gives a good approximation of negative common use. In the first convention, the windings
sequence reactance. are labeled H (high), L (low), M (medium); in the
The zero sequence reactance is much less than the second H (high), L (low), and T (tertiary). Unfortu-
others, producing a phase-to-ground fault current nately, the L winding in the second convention is
Phasors, Polarity, and Symmetrical Components 25

Figure 2-23 Equivalent positive, negative, and zero sequence circuits for some common and theoretical connections for two-
and three-winding transformers.

equivalent to M in the first. The tertiary winding or

voltage is generally the lowest.
On a common kVA base, the equivalent wye ZL ¼ ðZHL þ ZLT  ZHT Þ
leakage impedances are obtained from the following 2
equations: 1
ZH ¼ ðZHM þ ZHL  ZML Þ or
or 2
As a check, ZH plus ZM equals ZHM , and so on. The
ZH ¼ ðZHL þ ZHT  ZLT Þ wye is a mathematical equivalent valid for current and
2 voltage calculations external to the transformer bank.
1 The junction point of the wye has no physical
2 significance. One equivalent branch, usually ZM ðZL Þ,
26 Chapter 2

may be negative. On some autotransformers, ZH is impedance for a positive and negative sequence. Zero
negative. sequence impedance can vary from 2 to 6 times X1 ; a
The equivalent diagrams shown in Figure 2-23 are rough average for overhead lines is 3 to 3.5 times X1 .
satisfactory when calculations are to be made relative The resistance terms for the three sequences are
to one segment of a power system. However, a more usually neglected for overhead lines, except for lower-
complex representation is required when phase cur- voltage lines and cables. In the latter cases, line angles
rents and voltages are to be determined at points in the of 30 to 608 may exist, and resistance can be significant.
system having an intervening transformer between A good compromise is to use the impedance value
them and the point of discontinuity being examined. rather than reactance and neglect the angular differ-
For delta-wye transformers, a 308 phase shift must be ence in fault calculations. This yields a lower current to
accommodated. For ANSI standard transformers, the assure that the relay will be set sensitively enough.
high-voltage phase-to-ground voltage leads the low- Zero sequence mutual impedance resulting from
voltage phase-to-ground voltage by 308, irrespective of paralleled lines can be as high as 50 to 70% of the zero
which side the delta or wye is on. This phase shift may sequence self-impedance. This mutual impedance
be included in the equivalent per unit diagram by becomes an increasingly important factor as more
showing a 1 ff30 :1 ratio for it. lines are crowded into common rights of way.
The phase shift in the negative sequence network for
the delta-wye transformer is the same amount, but in
the opposite direction, to that in the positive sequence 5.5 Sequence Networks
network. The phase shift then, for an ANSI standard
transformer, would be 1 ff  30 :1 in the negative With the system assumed to be balanced or symme-
sequence per unit diagram. trical to the point of unbalance or fault, the three
The phase shift must be used in all the combinations sequence components are independent and do not
of Figure 2-23 where a wye and delta winding coexist. react with each other. Thus, three network diagrams
This effect is extremely important when consideration are required to separate the three sequence compo-
is being given to the behavior of devices on both sides nents for individual consideration: one for positive,
of such a transformer. one for negative, and one for zero sequence. These
sequence network diagrams consist of one phase to
neutral of the power system, showing all the compo-
5.4.3 Transmission Lines
nent parts relevant to the problem under considera-
For transmission lines, the positive and negative tion. Typical diagrams are illustrated in Figures 2-24
sequence reactances are the same. As a rule of thumb, through 2-26.
the 60-Hz reactance is roughly 0:8 O=mi for single The positive sequence network (Fig. 2-24) must
conductor overhead lines and 0:6 O=mi for bundled show both the generator voltages and impedances of
overhead lines. the generators, transformers, and lines. Balanced loads
The zero sequence impedance is always different may be shown from any bus to the neutral bus.
from the positive and negative sequence impedances. It Generally, however, balanced loads are neglected.
is a loop impedance (conductor plus earth and/or Compared to the system low-impedance high-angle
ground wire return), in contrast to the one-way quantities, they have a much higher impedance at a

Figure 2-24 Example system and positive sequence network.

Phasors, Polarity, and Symmetrical Components 27

A three-line system diagram is usually not required

to determine the zero sequence network, but if a
question arises as to the flow of zero sequence currents,
the three-line diagram can be useful. From this three-
phase system diagram, the zero sequence network
Figure 2-25 Negative sequence network for example requirements can be resolved by determining whether
system. or not equal and in-phase currents can exist in each of
the three phases. If the zero sequence current
component can flow, the zero sequence network must
reflect its path.
For simplicity, Figure 2-27 shows the generators
solidly grounded. In practice, however, solid ground-
ing is used only in very special cases.

Figure 2-26 Zero sequence network for example system.

5.6 Sequence Network Connections and

very low angle. In short, balanced loads complicate the
calculations and generally do not affect the fault The current flow direction and voltage connections
currents significantly. illustrated in Figure 2-28 must be followed for Eqs.
With two exceptions, the negative sequence network (2-29), (2-30), and (2-31) to apply. Current reference
(Fig. 2-25) will be a duplicate of the positive sequence direction in any circuit element must be the same in all
network: (1) There will be no generator voltages, since three networks to avoid confusion. Current flow in one
synchronous machines generate a positive sequence or more of the networks may reverse for some types of
only, and (2) the negative sequence reactance of unbalances, particularly if the networks are complex.
synchronous machinery may be different from the Reverse flow should be treated as a negative current to
positive, as previously described. For all practical ensure that it will be properly subtracted when
calculations involving faults or discontinuities remote determining the phase currents.
from the generating plant, however, X1 is assumed to Each sequence network is, of course, a per unit
be equal to X2 . diagram representing one of the three phases of the
The zero sequence network (Fig. 2-26) is quite symmetrical power system. Therefore, a resistor
different from the other two. First of all, it has no (reactor, impedance) connected between the system
voltage: Rotating machinery does not produce zero neutral and ground, as shown in Figure 2-28, must be
sequence voltage. Also, the transformer connections multiplied by 3 as indicated. In the system, 3I0 flows
require special consideration and grounding impe- through R; in the zero sequence network, however, I0
dances must be included. Figure 2-23 shows the zero flows through 3R, producing an equivalent voltage
sequence circuits for many transformers. drop.

Figure 2-27 Example system (generators shown solidly grounded for simplification).
28 Chapter 2

Figure 2-29 Three-phase fault and its network connection.

figures is the net impedance between the neutral bus

and selected fault location. Based on zero load, all
generated voltages (VAN ) are equal and in phase.
Since the three-phase fault is balanced, symmetrical
components are not required for this calculation.
However, since the positive sequence network repre-
sents the system, the network can be connected as
shown in Figure 2-29 to represent the fault.
For a phase-a-to-ground fault, the three networks
are connected in series (Fig. 2-30). Figure 2-31
illustrates a phase-b-c-to-ground fault and its sequence
network interconnection. The phase-b-to-phase-c fault
and its sequence connections are shown in Figure 2-32.
Fault studies normally include only three-phase
faults and single-phase-to-ground faults. Three-phase
faults are the most severe phase faults, whereas single-
phase-to-ground faults are the most common. Studies
of the latter faults provide useful information for
ground relaying.
A fundamental study of both series and shunt
unbalances was made by E. L. Harder in 1937. The
shunt unbalances summarized in Figure 2-33 are taken
from Harder’s study. Note that all the faults shown in
Figure 2-28 Sequence network connections and voltages. Figures 2-29 through 2-32 are also represented in
Figure 2-33.

5.7 Network Connections for Fault and General


The sequence networks can be interconnected at a

point of discontinuity, such as a fault. In such areas,
negative and zero sequence voltages are generated, as
previously described. Sequence network connections
for various types of common faults are shown in
Figures 2-29 through 2-32. From the three-phase
diagrams of the fault area, the sequence network
connections representing the fault can be derived.
These diagrams do not show fault impedance, and
fault studies do not include this effect except in very
rare cases. The single-sequence impedance Z1 ; Z2 ; Z0 Figure 2-30 Phase-to-ground fault and its sequence net-
(practically equivalent to X1 ; X2 ; X0 ) shown in the work connections.
Phasors, Polarity, and Symmetrical Components 29

positive, negative, and zero sequence interconnections

for the discontinuity shown in the top box are
illustrated in the three boxes below it.
Simultaneous faults require two sets of interconnec-
tions from either Figures 2-33 or 2-34 or both. As
shown in Figure 2-35, ideal or perfect transformers can
be used to isolate the two restrictions. Perfect
transformers are 100% efficient and have ratios of
1:1, 1:a, 1:a2.
It is sometimes necessary to use two transformers as
shown in Figure 2-35f. In this case, the first transfor-
mer (ratios 1 : ej30 , 1 : ej30 , and 1:1) represents the
wye-delta transformer, and the second transformer
(ratios 1:a2, 1:a, 1:1) represents the b-to-neutral fault.
These can be replaced by an equivalent transformer
with ratios 1 : ej150 , 1 : ej150 , and 1:1.
Figure 2-31 Double phase-to-ground fault and its sequence Figure 2-35a, for example, represents an open
network connections. phase-a conductor with a simultaneous fault to ground
on the x side. The sequence networks are connected for
In Figure 2-33, the entire symmetrical power system the open conductor according to Figure 2-34j, with
up to a point x of the shunt connection is represented by three 1:1 perfect transformers providing the restric-
a rectangular box. Inside the topmost box for each shunt tions required by Figure 2-33f. The manual calcula-
condition is a four-line representation of the shunt to be tions required, which involve the solution of
connected to the system at point x. The three lower boxes simultaneous equations, may be quite tedious.
for each shunt condition are the positive, negative, and
zero sequence representations of the shunt.
The sequence connections for the series unbalances,
such as open phases and unbalanced series impe- 5.8 Sequence Network Reduction
dances, are shown in Figure 2-34. As before, these
diagrams are taken from E. L. Harder’s study. Here When manual calculations are performed, the com-
again, the diagrams inside the topmost box for each plete system networks are reduced to the single
series condition represent the area under study, from impedance values of Figures 2-29 through 2-32.
point x on the diagrams left to point y on the right. The To simplify this reduction, with negligible effect on
power system represented by the box is open between x the results, the following basic assumptions are some-
and y to insert the circuits shown inside the box. Points times made:
x and y can be any distance apart, as long as there is no All generated voltages are equal and in phase.
tap or other system connection between them. The All resistance is neglected, or the reactance of
machines and transformers is added directly
with line impedances.
All shunt reactances are neglected, including loads,
charging, and magnetizing reactances.
All mutual reactances are neglected, except on
parallel lines.
By using these assumptions, the positive sequence
network can be drawn with a single-source voltage Van
connected to the generator impedances by a bus.
If voltages are different, either the voltages must be
retained in the network or Thevenin theorem and
superposition must be used to reduce the network and
Figure 2-32 Phase-to-phase fault and its sequence network calculate fault currents and voltages. Note that for the
connections. series unbalances of Figure 2-34, a difference in
30 Chapter 2

Figure 2-33 Sequence network interconnections for shunt balanced and unbalanced conditions.

voltage—either magnitude, phase angle, or both—is per unit current flows in these single-sequence impe-
required for current to flow. dances at the fault or point of discontinuity.
The single-sequence impedances Z1, Z2 and Z0 of Network reduction calculations for the system of
Figures 2-29 through 2-32 will be different for each Figure 2-24 are illustrated in Figures 2-36, 2-37, and 2-
fault location because of the different network reduc- 38. In these figures, X1, X2, and X0 are the impedances
tions. During the network reduction, the distribution of between the neutral bus and the fault at bus G. I1R, I1L,
currents in the various branches should be calculated, I2R, I2L, I0R, I0L are the per unit distribution factors. I1,
both as a check and to determine the current flow I2, and I0 are all assumed to be equal to 1 per unit.
through the relays involved in a fault. These distribu- Analog or digital studies should be tailored to
tion factors are calculated with the assumption that 1 produce outputs that allow each branch current in
Phasors, Polarity, and Symmetrical Components 31

Figure 2-34 Sequence network interconnections for series balanced and unbalanced conditions.

each network to be identified. For single-phase-to- correct relay types and settings to be selected in a
ground faults, 3I0 is required for relays. minimal amount of time for the entire power system.
When using the computer for sequence network The following steps must be performed for calculat-
reduction, the impedance data are input for the ing fault currents and voltages:
positive and zero sequence networks, along with bus
and fault node points. The network is then solved for Obtain a complete single-line diagram for the entire
three-phase and single-phase-to-ground faults. Tabu- system, including generators, transformers, and
lated printed data are provided for phase-a fault transmission lines, along with the positive,
current and three-phase fault voltages, along with the negative, and zero sequence impedances for
corresponding 3I0, 3V0 values for the phase-to-ground each component.
fault. I2 and V2 values should also be obtained for Prepare a single-line impedance diagram from the
negative sequence relays. system diagram or establish the nodes in a digital
These voltage and current values are needed for not study for the positive, negative, and zero
only faults near the relay, but also those several buses or sequence networks.
lines away. Among the operating conditions normally Reduce the impedance values of all network
considered are maximum and minimum generation, branches to a common base. Values may be
selected lines out of service, and line-end faults where the expressed as per unit on a common kVA base or
adjacent breaker is open. This information allows the as ohms impedance on a common voltage base.
32 Chapter 2

Figure 2-35 Representations for simultaneous unbalances.

Obtain, or have the computer obtain, the equivalent seldom of use as relays generally see a fraction of
single impedance of each sequence network, that current except for radial circuits.
current distribution factors, and equivalent source
voltage for the positive-phase sequence network.
All quantities must be referred to the proper base. 5.9 Example of Fault Calculation on a Loop-
Interconnect the networks or utilize the computer Type Power System
program to represent the fault type involved, and
calculate the total fault current at the fault. For the typical loop system shown in Figure 2-39, the
Determine the current distribution and voltages as generator units at stations D, S, or E could each be
required in the system. Total fault current is combinations of several machines. Alternatively, they
Phasors, Polarity, and Symmetrical Components 33

Figure 2-36 Network reduction for example system

(Figure 2-24) fault at bus G.

Figure 2-38 Final network reduction for fault at bus G in

Figure 2-24.

loops, at least one must be converted to wye-equivalent

in order to reduce the networks. After one loop is
chosen (arbitrarily), the equivalent X, Y, Z branches
for an equivalent wye are dotted in as shown in Figures
2-40 and 2-41.
The X, Y, Z conversion from delta to wye-
equivalent is a simple process: The X branch of the
wye-equivalent is the product of the two delta
reactances on either side divided by the sum of the
three delta impedances. The same relation applies to

Figure 2-37 Network reduction and current distribution.

could represent the equivalent of a complex system up

to the bus. All the impedances have been reduced to a
common base, as indicated in the diagram. The
positive sequence network for this system is shown in
Figure 2-40, the zero sequence network in Figure 2-41.
The negative sequence network is equal to Figure 2-40,
except that Van is not present.
To perform this sample calculation of a phase-to-
ground fault on the bus at station D, the networks
must be reduced to a single reactance value between Figure 2-39 Single line diagram for a typical loop-type
the neutral bus and fault point. Of the several delta power system.
34 Chapter 2

Figure 2-41 Zero sequence network reduction for the

system of Figure 2-39.

Figure 2-40 Positive sequence network reduction for the

system of Figure 2-39.

The networks now reduce to the simpler forms shown

the Y and Z branches. Thus, in Figures 2-40 and 2-41, in Figure 2-40c. Since the two upper branches of
the networks are reduced as follows: each network are in parallel, they can be reduced as

Positive and negative Zero sequence

sequence networks network Positive and negative Zero sequence
sequence networks network
X1 ¼ 24628 ¼ j10:84 X0 ¼ 966 ¼ j0:72
62 75
0:4716 0:5284 0:2594 0:7406
Y1 ¼ 28610 ¼ j4:52 Y0 ¼ 6660 ¼ j4:8
62 75 44:52649:87 56:86162:2
24610 9660 94:39 219:0
Z1 ¼ ¼ j3:87 Z0 ¼ ¼ j7:2
62 75 ¼ 23:52 ¼ 42:07
Phasors, Polarity, and Symmetrical Components 35

These reductions are shown in Figures 2-40d and

2-41c. The remaining branches are in parallel and can
also be reduced:

Positive and negative

sequence networks Zero sequence network

0:4621 0:5379 0:8106 0:1894

X1 ¼ X2 ¼ 34:36640 X0 ¼ 42:79610
74:36 52:79
¼ j18:48% ¼ j8:11%

The numbers written above the equations are the

distribution factors for the parallel circuits. These
factors are expressed as the ratio of each term in the
numerator and denominator. Determining these fac-
tors provides a convenient check on the calculations,
since the sum of the two fractions must be 1.
Distribution factors can be determined by working
back through the reduction. The factors should be
written on the diagrams as shown in Figure 2-42.
The distribution factors for the upper parallel Figure 2-42 Per unit current distribution for AG fault at D.
branches of Figure 2-40c are determined as follows:

Positive sequence network Positive sequence network

44:52% branch : 0:528460:5379 ¼ 0:2842 0:53796j10:84 þ 0:28426j4:52
¼ 0:2541
49:87% branch : 0:471660:5379 ¼ 0:2537
0:53796j10:84 þ 0:25376j3:87
0:5379 ðcheckÞ ¼ 0:2838
0:2537 þ j3:87 þ 0:28426j4:52
The distribution factors in the zero sequence ¼ 0:0301
network are

Zero sequence network Zero sequence network

162:2% branch : 0:259460:1894 ¼ 0:0491 0:18946j0:72 þ 0:14036j4:52
¼ 0:1350
56:8% branch : 0:740660:1895 ¼ 0:1403 j6
0:1894 ðcheckÞ 0:18946j0:72 þ 0:04916j7:20
¼ 0:0544
In turn, these distribution factors are added to the 0:0491 þ j7:20 þ 0:14036j4:80
¼ 0:0053
diagram, as shown in Figure 2-42b. j60
The delta current distribution factors are obtained
from the X, Y, Z equivalents. The conversion Figure 2-42 shows the complete per unit distribution
technique is straightforward: The voltage drop across for the original network of Figure 2-39.
two of the wye branches is equivalent to the drop The three networks are connected in series for the
across the delta branch. Calculating from Figure 2-40c, phase-to-ground fault (Fig. 2-28). For convenience, the
we obtain sequence currents are calculated in per unit values:
36 Chapter 2

I1 ¼ I2 ¼ I0 From Figure 2-42, first the sequence and phase

j1:0 voltages are calculated at bus S:
j0:1848 þ j0:1848 þ j0:0811
1:0 V1 ¼ j1:0  0:62976j0:24
¼ ¼ j1:0  j0:1511
¼ 2:22 p:u: ¼ j0:8489 p:u: ð53;912:39 VÞ
V2 ¼ 0  0:62976j0:24
The 100% (1 p.u. base) current is
¼ j0:1511 p:u: ð9596:14 VÞ
kVA base
IB ¼ pffiffiffi V0 ¼ 0  0:12076j0:09
3 kV
¼ j0:0109 p:u: ð692:24 VÞ
¼ pffiffiffi Vag ¼ V1 þ V2 þ V0
¼ j0:6869 p:u: ð43;624:01 VÞ
¼ 524:86 A at 110 kV
I1 ¼ I2 ¼ I0 Vbg ¼ a2 V1 þ aV2 þ V0
¼ 2:226524:86 ¼ 0:8489ff  30 þ 0:1511ff þ 30  j0:0109
¼ 1164:55 A at 110 kV ¼ 0:7352  j0:4245 þ 0:1309 þ j0:0756  j0:0109
¼ 0:8661  j0:3598
The current flowing in each branch of the networks can
¼ 0:9379ff  22:56 p:u: ð59;594:65 VÞ
now be determined by multiplying the actual fault cur-
rent by the distribution factor. These currents may be Vcg ¼ aV1 þ a2 V2 þ V0
expressed in either per unit or ampere values. Currents ¼ 0:8489ff210 þ 0:1511ff150  j0:0109
in the fault are calculated for each phase as follows: ¼ 0:7352  j0:4245  0:1309 þ j0:0756  j0:0109
Ia ¼ 3I1 ¼ 3I2 ¼ 3I0 ¼ 6:66 p:u: ¼ 0:8661  0:3598
or ¼ 0:9379ff202:56 p:u: ð59;564:65 VÞ

Ia ¼ 3493:66 A110 kV Next, the sequence and phase voltages are calculated at
bus D, the fault location:
Ib ¼ ða I1 þ aI2 þ I0 Þ

¼ ðI1 þ I0 Þ ¼ 0 V1 ¼ j1:0  1:02536j0:40

Ic ¼ ðaI1 þ a I2 þ I0 Þ
¼ j1:0  j0:4101
¼ ðI1 þ I0 Þ ¼ 0 ¼ j0:5899 p:u: ð37;463:68 VÞ
For each branch, the per unit positive, negative, and V2 ¼ 0  1:02536j0:40
zero sequence currents can then be used to determine ¼ j0:4101 p:u: ð26;044:85 VÞ
the individual phase currents by using Eqs. (2-23), V0 ¼ 0  1:79866j0:1
(2-24), and (2-25). These are recorded in Figure 2-43. ¼ j0:1798 p:u: ð11;418:83 VÞ
Next, the sequence and phase voltages at each bus are
Vag ¼0
determined as in Figure 2-28. It is convenient to calculate
the voltages in per unit values. Note that the impedances Vbg ¼ 0:5899ff  30 þ 0:4101ff30  j0:1798
listed in Figure 2-39 appear in percent, rather than ¼ 0:5109  j0:2950 þ 0:3552 þ j0:2051  j0:1798
ohms, and may be converted easily to per unit. ¼ 0:8661  j0:2697
In the following calculations, the values in parenth-
eses are volts, converted from the per unit values for ¼ 0:9071ff  17:30 p:u: ð57;608:59 VÞ
the 110-kV system of Figure 2-39: Vcg ¼ 0:5899ff210 þ 0:4101ff150  j0:1798
¼ 0:5109  j0:2950  0:3552 þ j0:2051  j0:1798
Vline-to-neutral ¼ 1:0 p:u:
¼ 0:8661  j0:2697
110;000 V
¼ pffiffiffi ¼ 0:9071ff197:30 p:u: ð57;608:59 VÞ
¼ 63;508:53 V Similarly, the sequence and phase voltages can be
Phasors, Polarity, and Symmetrical Components 37

Figure 2-43 Current and voltage distribution for a single phase-to-ground fault at bus ‘‘D’’ of the system of Figure 2-37.

calculated at bus E: 5.10 Phase Shifts Through Transformer Banks

Vag ¼ j0:6352 p:u: ð40;340:62 VÞ In these fault calculations, the phase shifts through the

Vbg ¼ 0:9502ff  24:30 p:u: ð60;345:80 VÞ wye-delta transformer banks were not considered. In
this example, only a 110-kV system fault, with its
Vcg ¼ 0:9502ff204:30 p:u: ð60;345:80 VÞ currents and voltages, was involved. The effect of the
phase shift through the transformer banks could not,
Finally, the voltages are calculated at bus R: however, have been neglected if currents and voltages
were required for the opposite side of the power
Vag ¼ j0:3646 p:u: ð23;155:21 VÞ transformers.
Vbg ¼ 0:8909ff13:59 p:u: ð56;579:75 VÞ If the transformer bank is wye-connected on the
Vcg ¼ 0:8909ff193:59 p:u: ð56;579:75 VÞ high-voltage side, as shown in Figure 2-44, the general
equations for one phase are
The sequence voltages calculated above, as shown in IA ¼ nðIa  Ic Þ ð2-30Þ
Figure 2-43, complete the analysis of the single-phase- Van ¼ nðVAn  VBn Þ ¼ nVAB ð2-31Þ
to-ground fault at bus D in the system of Figure 2-39.
All the distributed current and voltage values for the The lowercase subscripts represent high-side quantities
system are displayed in Figure 2-43. and the capital letter subscripts low-side quantities. In
38 Chapter 2

Va2 ¼ nðVA2  VB2 Þ

¼ nðVA2  aVA2 Þ
¼ nð1  aÞVA2
¼ n 3VA2 ff  30
Va2 ¼ NVA2 ff  30
VA2 ¼ ff30
If a power transformer bank is connected delta on the
high-voltage side, as shown in Figure 2-45, the general
equations for one phase are
Figure 2-44 Connections and phasors for an ANSI
standard power transformer bank with the wye connection Ia ¼ ðIA  IB Þ ð2-38Þ
on the high side (Van leads VAN by 308). n
VA ¼ ðVa  Vc Þ ð2-39Þ
the balanced or symmetrical transformer bank, the
sequences are independent.
Applying only positive sequence quantities to Eqs.
Consequently, positive sequence only is first applied
(2-38) and (2-39),
to Eqs. (2-30) and (2-31):

IA1 ¼ nðIa1  Ic1 Þ 1

Ia1 ¼ ðIA1  IB1 Þ
¼ nðIa1  aIa1 Þ n
¼ nð1  aÞIa1 1
¼ ðIA1  a2 IA1 Þ
pffiffiffi n
¼ n 3Ia1 ff  30 1
¼ ð1  a2 ÞIA1
IA1 ¼ NIA1 ff  30 ð2-32Þ n
IA1 3IA1
Ia1 ¼ ff30 ¼ ff30
N n
Va1 ¼ nðVA1  VB1 Þ IA1
Ia1 ¼ ff30 ð2-40Þ
¼ nðVA1  a2 VA1 Þ ð2-33Þ N
IA1 ¼ NIa1 ff  30 ð2-41Þ
¼ nð1  a2 ÞVA1
Va1 ¼ n 3VA1 ff30
Va1 ¼ NVA1 ff30 ð2-34Þ
VA1 ¼ ff  30 ð2-35Þ

Next, only negative sequence quantities are applied to

Eqs. (2-30) and (2-31):

IA2 ¼ nðIa2  Ic2 Þ

¼ nðIa2  a2 Ia2 Þ
¼ nð1  a2 ÞIa2
¼ n 3Ia2 ff30
IA2 ¼ NIa2 ff30 ð2-36Þ
Figure 2-45 Connections and phasors for an ANSI
Ia2 ¼ ff  30 ð2-37Þ standard power transformer bank with the delta connection
N on the high side (Van leads VAN by 308).
Phasors, Polarity, and Symmetrical Components 39

VA1 ¼ ðVa1  Vc1 Þ Table 2-2 Phase Shift Relations for Power Transformer
n Banks
¼ ðVa1  aVa1 Þ High side in terms Low side in terms
1 of low sidea of high sidea
¼ ð1  aÞVa1
pffiffiffi IA1
Ia1 ¼ ff30 IA1 ¼ NIa1 ff  30
3Va1 N
¼ ff  30 V
n Va1 ¼ NVA1 ff30 VA1 ¼ a1 ff  30
Va1 I
Ia2 ¼ A2 ff  30 IA2 ¼ NIa2 ff30
VA1 ¼ ff  30 ð2-42Þ N
Va1 ¼ NVA1 ff30 ð2-43Þ Va2 ¼ NVA2 ff  30 VA2 ¼ A2 ff30
Then, applying only negative sequence quantities to The lowercase subscripts represent high-side quantities, and the
capital letter subscripts low-side quantities.
Eqs. (2-38) and (2-39), we obtain
Ia2 ¼ ðIA2  IB2 Þ
1 5.11 Fault Evaluations
¼ ðIA2  aIA2 Þ
1 The sample calculation of a phase-to-ground fault on a
¼ ð1  aÞIA2
pffiffiffi loop system (see Sec. 5.9) was made at no load; that is,
3IA2 before the fault all currents throughout the system
¼ ff  30 were zero.
IA2 With a ground fault, current flows in not only the
Ia2 ¼ ff  30 ð2-44Þ faulted phase ‘‘a,’’ but also the unfaulted ‘‘b’’ and ‘‘c’’
IA2 ¼ NIa2 ff30 ð2-45Þ phases. The positive and zero sequence distribution
factors on any loop system will be different. Conse-
VA2 ¼ ðVa2  Vc2 Þ quently, the positive, negative, and zero sequence
n currents will not add up to zero in the unfaulted
¼ ðVa2  a2 Va2 Þ phases. On a radial system (one with a source at one
n end only for both the positive and zero sequences), the
¼ ð1  a2 ÞVa2 three network distribution factors will all be equal to 1.
pffiffiffi For a phase-a-to-ground fault on these circuits, Ib
3Va2 equals Ic , which equals 0.
¼ ff30
n In practice, only 3I0 and related 3V0 ; V2 , and I2
Va2 values would be recorded for a phase-to-ground fault.
VA2 ¼ ff30 ð2-46Þ
N The phase currents and voltages shown in Figure 2-43
Va2 ¼ NVA2 ff  30 ð2-47Þ were provided for academic purposes.
The reason for showing 3I0 , rather than the faulted
If the bank is connected according to ANSI phase current, can be seen from Figure 2-43. In most
standards, the formulas are the same and not depen- circuits, there is a significant difference between the Ia
dent on whether the wye or the delta is on the high side. and 3I0 currents in any loop network. In a radial
In either case, the positive sequence quantities are system, however, Ia is equal to 3I0 and ground relays
shifted 308 in one direction, while the negative sequence operate on 3I0 .
quantities are shifted 308 in the opposite direction. On phase-to-ground faults, the phase relays will
These relations for ANSI standard connections are receive current and may start to operate. Coordination
summarized in Table 2-2. Zero sequence quantities are between ground and phase relays is usually not
not affected by phase shift. These either pass directly necessary. The principal reason there are so few
through the bank or, more commonly, are blocked by coordination problems is that phase relays must be
the connections. Thus, in a wye-delta bank, zero set above load (5 A secondary), whereas ground relays
sequence current and voltage on one side cannot pass are conventionally set at 0.5 to 1.0 A secondary. Since
through the bank to the other side. the ground relays are more sensitive, they will generally
40 Chapter 2

not miscoordinate with the phase relays. If higher converter of positive sequence into negative sequence
ground settings are used, the likelihood of miscoordi- and, for ground faults, into zero sequence current.
nation is increased. This is illustrated by a voltage plot for various faults
Under any fault condition, the total current flowing on a simple system (Fig. 2-46). For simplicity, assume
into the ground must equal the total current flowing up Z1 equals Z2 equals Z0 . During faults, the voltage
the neutrals. With an autotransformer, however, inside the generators does not change unless the fault
current can flow down the neutral. In this case, the persists long enough for the internal flux to change. No
fault current plus the autotransformer neutral current appreciable voltage change should occur in high- or
equals the current up the other transformer neutrals. medium-speed relaying.
The convention that current flows up the neutral For a solid three-phase fault, the voltage at the fault
when current is flowing down into the earth at the fault is zero. Therefore, high positive sequence-phase
has given rise to the idea that the grounded wye-delta currents flow to produce the gradient shown in the
transformer bank is a ground source, a source of zero plot of Figure 2-46. For a phase-to-phase fault,
sequence current. This long-established idea is not, in negative sequence voltage is produced by the fault
fact, correct. The fault is the true source. It is a itself. Negative sequence current, then, flows through-

Figure 2-46 Voltage gradient for various types of faults.

Phasors, Polarity, and Symmetrical Components 41

out the system. The same general conditions also apply normal third phase-to-neutral voltage. Two of the
to phase-to-ground faults, except that since Va is zero, phase currents have a large lagging increase.
V2 and V0 are negative. For a single-phase-to-ground fault, on the other
In summary, the positive sequence voltage is always hand, one phase-to-neutral voltage is collapsed relative
highest at the generators or sources and lowest at a to the other two phases. Similarly, one phase current
fault. In contrast, negative and zero sequence voltages has a large value and lags the line-to-ground voltage.
are always highest at the fault and lowest at the With wye-delta transformers between the fault and
‘‘sources.’’ measurement point, the positive sequence quantities
The phasor diagrams of Figure 2-20 illustrate the shift 308 in one direction, and the negative sequence
same phenomena, from a different viewpoint. In a quantities shift 308 in the opposite direction. As a
three-phase fault, the voltages collapse symmetrically, result, a phase-to-ground fault on the wye side of a
except inside the generator. The three currents have bank has the appearance of a phase-to-phase fault on
a large symmetrical increase and lagging shift of the delta side.
angle. Figures 2-47 and 2-48 offer a final look at sequence
Other phase faults shown in Figure 2-20 are currents and voltages for faults. Note that the positive
characterized by the relative collapse of two of the sequence currents and voltages, shown in the left-hand
phase-to-neutral voltages, compared to the relatively columns, have approximately the same phase relations

Figure 2-47 Sequence currents for various faults. Assumes Figure 2-48 Sequence voltages for various faults. Assumes
Z1 ¼ Z2 ¼ Z0 . Z1 ¼ Z2 ¼ Z0 .
42 Chapter 2

for all types of faults. At the fault are various A zero sequence ð3I0 Þ current filter is obtained by
nonsymmetrical currents and voltages, as shown in connecting three current transformers in parallel. A
the far right-hand column. The negative and, some- zero sequence ð3V0 Þ voltage filter is provided by the
times, the zero sequence quantities provide the transi- wye-grounded-broken-delta connection for a voltage
tion between the symmetrical left-hand column and transformer or an auxiliary. Positive and negative
nonsymmetrical right-hand column. These quantities sequence current and voltage filters are described in
rotate and change to produce the nonsymmetrical, or Chapter 3.
unbalanced, quantity when added to the positive
These phasors can be constructed easily by
remembering which fault quantity should be mini-
mum or maximum. In a phase c-a fault, for example,
phase-b current will be small. Thus, Ib2 will tend to Table 2-3 Protective Relays Using Symmetrical
be opposite Ib1 . Since phase-b voltage will be Component Quantities for Their Operation
relatively uncollapsed, Vb1 and Vb2 will tend to be Sequence quantities
in phase. After one sequence phasor is established, Device no. Application used
the others can be derived from Eq. (2-12) and
Figure 2-21. 50N, 51N Ground overcurrent I0
59N Ground voltage V0
67N Ground directional I0 with I0 or V0I0
overcurrent or V2I2
6 SYMMETRICAL COMPONENTS AND 32N Ground product I20 or I0, V0
RELAYING overcurrent
21N Ground distance I0 , V 0
Since ground relays operate from zero sequence I0, V0, V1 þ V2
quantities, all ground relay types use symmetrical 87 Phase and ground pilot K1I1 þ K2I2 þ K0I0
components. A number of other protective relays use 46 Phase unbalance voltage V2
46 Phase unbalance current I2
combinations of the sequence quantities, as summar-
Blown fuse detection V0 and not I0
ized in Table 2-3.
Basic Relay Units
Revised by: W. A. ELMORE

1 INTRODUCTION current or voltage applied to the coil exceeds the

pickup value, the plunger moves upward to operate a
Protective relays for power systems are made up of one set of contacts. The force F which moves the plunger is
or more fault-detecting or decision units, along with proportional to the square of the current in the coil.
any necessary logic networks and auxiliary units. The plunger unit’s operating characteristics are
Because a number of these fault-detecting or decision largely determined by the plunger shape, internal
units are used in a variety of relays, they are called core, magnetic structure, coil design, and magnetic
basic units. Basic units fall into several categories: shunts. Plunger units are instantaneous in that no
electromechanical units, sequence networks, solid-state delay is purposely introduced. Typical operating times
units, integrated circuits, and microprocessor architec- are 5 to 50 msec, with the longer times occurring near
ture. Combinations of units are then used to form the threshold values of pickup.
basic logic circuits applicable to protective relays. The unit shown in Figure 3-1a is used as a high-
dropout instantaneous overcurrent unit. The steel
plunger floats in an air gap provided by a nonmag-
netic ring in the center of the magnetic core. When the
coil is energized, the plunger assembly moves upward,
Four types of electromechanical units are widely used:
carrying a silver disk that bridges three stationary
magnetic attraction, magnetic induction, D’Arsonval,
contacts (only two are shown). A helical spring
and thermal units.
absorbs the ac plunger vibrations, producing good
contact action. The air gap provides a ratio of
2.1 Magnetic Attraction Units
dropout to pickup of 90% or greater over a two-to-
one pickup range. The pickup range can be varied
Three types of magnetic attraction units are in
from a two-to-one to a four-to-one range by the
common use: plunger (solenoid), clapper, and polar.
adjusting core screw. When the pickup range is
The plunger unit, shown in Figure 3-1, is typically used
increased to four to one, the dropout ratio will
in SC, SV, and ITH relays; the clapper-type unit (Fig.
decrease to approximately 45%.
3-2) in SG, AR, ICS, IIT, and MG relays; and the
The more complex plunger unit shown in Figure
polar-type unit (Fig. 3-3) in HCB, HU, and PM-type
3-1b is used as an instantaneous overcurrent or voltage
unit. An adjustable flux shunt permits more precise
settings over the nominal four-to-one pickup range.
2.1.1 Plunger Units
This unit is relatively independent of frequency,
Plunger units have cylindrical coils with an external operating on dc, 25-Hz, or nominal 60-Hz frequency.
magnetic structure and a center plunger. When the It is available in high- and low-dropout versions.

44 Chapter 3

Figure 3-1 Plunger-type units.

2.1.2 Clapper Units

Clapper units have a U-shaped magnetic frame with a
movable armature across the open end. The armature
is hinged at one side and spring-restrained at the other.
When the associated electrical coil is energized, the
armature moves toward the magnetic core, opening or
closing a set of contacts with a torque proportional to
the square of the coil current. The pickup and dropout
values of clapper units are less accurate than those of
plunger units. Clapper units are primarily applied as
auxiliary or go/no-go units.
Four clapper units are shown in Figure 3-2. Those
illustrated in Figures 3-2a and 3-2b have the same
general design, but the first is for dc service and the
second for ac operation. In both units, upward
movement of the armature releases a target, which
drops to provide a visual indication of operation (the
target must be reset manually). The dc ICS unit (Fig.
3-2a) is commonly used to provide a seal-in around the
main protective relay contacts. The ac IIT unit (Fig.
3-2b) operates as an instantaneous overcurrent or
instantaneous trip unit. It is equipped with a lag-loop
to smooth the force variations due to the alternating
current input. Its adjustable core provides pickup
adjustment over a nominal four-to-one range.
The SG (Fig. 3-2c) and MG clapper units provide a
wide range of contact multiplier auxiliaries: The SG
has provisions for four contacts (two make and two
break), and the MG will accept six. The AR clapper
unit (Fig. 3-2d) operates in 2 to 4 msec, with four
Figure 3-2 Four clapper units. contacts suitable for breaker tripping.
Basic Relay Units 45

overcome this tendency and the contact will move to

the left. Depending on design and adjustments, this
polarizing action can be gradual or quick. The left-gap
adjustment (Fig. 3-3b) controls the pickup value, the
right-gap adjustment the reset value. Some units use
both an operating and a restraining coil on the
armature. The polarity of the restraint coil tends to
maintain the contacts in their initial position. Current
of sufficient magnitude applied to the operating coil
will provide a force to overcome the restraint, causing
the contacts to change position. A combination of
normally open or normally closed contacts is available.
These polar units operate on alternating current
through a full-wave rectifier and provide very sensitive,
high-speed operation on very low energy levels.
The operating equation of the polar unit is
K1 Iop  K2 Ir ¼ ð3-1Þ
where K1 and K2 are adjusted by the magnetic shunts;
K3 is a design constant; f is the permanent magnetic
flux; Iop is the operating current; and Ir is the restraint
current in milliamperes.
Figure 3-3 Polar-type unit 2.2 Magnetic Induction Units

There are two general types of magnetic induction

2.1.3 Polar Units units: induction disc and cylinder units. The induction
disc unit (Fig. 3-4) is typically used in CO, CV, CR,
Polar units (Fig. 3-3) operate from direct current IRV, IRD, CW, CA, and CM relays. The cylinder unit
applied to a coil wound around the hinged armature in (see Fig. 3-6) is most commonly used in KD-line, KC,
the center of the magnetic structure. A permanent KDXG, KF, KRD, KRC, and KRP relays.
magnet across the structure polarizes the armature-gap
2.2.1 Induction Disc Units
poles, as shown. The nonmagnetic spacers, located at
the rear of the magnetic frame, are bridged by two Originally, induction disc units were based on the
adjustable magnetic shunts. This arrangement enables watthour meter design. Modern units, however,
the magnetic flux paths to be adjusted for pickup and although using the same operating principles are quite
contact action. With balanced air gaps (Fig. 3-3a), the different. All operate by torque derived from the
flux paths are as shown and the armature will float in interaction of fluxes produced by an electromagnet
the center with the coil deenergized. With the gaps with those from induced currents in the plane of a
unbalanced (Fig. 3-3b), some of the flux is shunted rotatable aluminum disc. The E unit in Figure 3-4a has
through the armature. The resulting polarization holds three poles on one side of the disc and a common
the armature against one pole with the coil deener- magnetic member or ‘‘keeper’’ on the opposite side.
gized. The coil is arranged so that its magnetic axis is in The main coil is on the center leg. Current I in the main
line with the armature and at a right angle to the coil produces flux, which passes through the air gap
permanent magnet axis. Current in the coil magnetizes and disc to the keeper. (A small portion of the flux is
the armature either north or south, increasing or shunted off through the side air gap.) The flux fT
decreasing any prior polarization of the armature. If, returns as fL through the left-hand leg and fR through
as shown in Figure 3-3b, the magnetic shunt adjust- the right-hand leg, where fT ¼ fL þ fR . A short-
ment normally makes the armature a north pole, it will circuited lagging coil on the left leg causes fL to lag
move to the right. Direct current in the operating coil, both fR and fT , producing a split-phase motor action.
which tends to make the contact end a south pole, will (The phasors are shown in Figure 3-5.)
46 Chapter 3

Figure 3-5 Phasors and operations of the ‘‘E’’ unit

induction disc.

the total flux produced by main coil current I. The

three fluxes cross the disc air gap and induce eddy
currents in the disc. These eddy currents react with the
pole fluxes and produce the torque that rotates the
disc. With the same reference direction for the three
fluxes as shown in Figure 3-5b, the flux shifts from left
to right and rotates the disc clockwise, as viewed from
the top.
There are many alternative versions of the induction
disc unit. The unit shown in Figure 3-4, for example,
may have a single current or voltage input. The disc
always moves in the same direction, regardless of the
direction of the input. If the lag coil is open, no torque
will exist. Other units can thus control torque in the
induction disc unit. Most commonly, a directional unit
is connected in the lag coil circuit. When the
directional unit’s contact is closed, the induction disc
unit has torque; when the contact is open, the unit has
no torque.
Induction disc units are used in power or directional
applications by substituting an additional input coil for
the lag coil in the E unit. The phase relation between
the two inputs determines the direction of the
operating torque.
A spiral spring on the disc shaft conducts current to
Figure 3-4 Induction disc unit. the moving contact. This spring, together with the
shape of the disc (an Archimedes spiral) and design of
the electromagnet, provides a constant minimum
Flux fL induces voltage Vs , and current Is flows, operating current over the contact travel range. A
essentially in phase, in the shorted lag coil. Flux fT is permanent magnet with adjustable keeper (shunt)
Basic Relay Units 47

dampens the disc, and magnetic plugs in the electro- attached to the top of the cylinder and the stationary
magnet control the degree of saturation. The spring contacts. A spiral spring provides reset torque.
tension, damping magnet, and magnetic plugs allow Operating torque is a function of the product of the
separate and relatively independent adjustment of the two operating quantities applied to the coils wound on
unit’s inverse-time current characteristics. the four poles of the electromagnet and the sine of the
angle between them. The torque equation is

2.2.2 Cylinder Units T ¼ KI1 I2 sin f12  Ks ð3-2Þ

The operation of a cylinder unit is similar to that of an where K is a design constant; I1 and I2 are the currents
induction motor with salient poles for the stator through the two coils; f12 is the angle between I1 and
windings. Shown in Figure 3-6 , the basic unit used I2 ; and Ks is the restraining spring torque. Different
for relays has an inner steel core at the center of the combinations of input quantities can be used for
square electromagnet, with a thin-walled aluminum different applications, system voltages or currents, or
cylinder rotating in the air gap. Cylinder travel is network voltages.
limited to a few degrees by the moving contact
2.3. D’Arsonval Units

In the D’Arsonval unit, shown in Figure 3-7, a

magnetic structure and an inner permanent magnet
form a two-pole cylindrical core. A moving coil loop in
the air gap is energized by direct current, which reacts
with the air gap flux to create rotational torque. The
D’Arsonval unit operates on very low energy input,
such as that available from dc shunts, bridge networks,
or rectified ac. The unit can also be used as a dc
contact-making milliammeter or millivoltmeter.

2.4 Thermal Units

Thermal units consist of bimetallic strips or coils that

have one end fixed and the other end free. As the
temperature changes, the different coefficients of
thermal expansion of the two metals cause the free
end of the coil or strip to move. A contact attached to
the free end will then operate based on temperature


Static networks with three-phase current or voltage

inputs can provide a single-phase output proportional
to positive, negative, or zero sequence quantities.
These networks, also known as sequence filters, are
widely used.

3.1 Zero Sequence Networks

In zero sequence networks, three current transformer

Figure 3-6 Cylinder unit. secondaries, connected in parallel, provide 3I0 from Ia ,
48 Chapter 3

VF is obtained from input currents Ia , Ib , and Ic , with

neutral ð3I0 Þ return. By using Thevenin’s theorem,
these three-phase networks can be reduced to a simple
equivalent circuit, as shown in Figure 3-8b. VF is the
open circuit voltage at the output, and Z the
impedance looking back into the three-phase network.
Zs is the self-impedance of the three-winding reactor’s
secondary with mutual impedance Xm .
The open circuit voltage (Fig. 3-8a) with switch r
open and switch s closed is the drop from VF ðþÞ to
VF ðÞ.

VF ¼ jðIc  Ib ÞXm þ Ia R1 þ 3I0 R0 ð3-3Þ

From the basic symmetrical component equations [Eq.

(2-24) and (2-25) in Chapter 2], we have
pffiffiffi pffiffiffi
Ic  Ib ¼ j 3I1  j 3I2 ð3-4Þ

Substituting this and the sequence equation (2-23) for

Ia in Eq. (3-3),
pffiffiffi pffiffiffi
VF ¼  3Xm I1 þ 3Xm I2 þ R1 I1
þ R1 I2 þ R1 I0 þ 3I0 R0
pffiffiffi pffiffiffi ð3-5Þ
¼ ðR1  3Xm ÞI1 þ ðR1 þ 3Xm ÞI2
þ ðR1 þ 3R0 ÞI0

Varying Xm , R1 , R0 , and the connections produces

different output characteristics. In some applications,
the currents Ib and Ic are interchanged, changing

Figure 3-7 D’Arsonval-type unit.

Ib , and Ic inputs. Similarly, the secondaries of three-

phase voltage transformers, connected in series with
the primary in grounded wye, provide 3V0 .

3.2 Composite Sequence Current Networks

The network shown in Figure 3-8 can be adapted for a

variety of single-phase outputs. Output filter voltage Figure 3-8 Composite sequence current network.
Basic Relay Units 49

Table 3-1 Typical Sequence Network Combinations

VF reduces from:

Network type Switch r Switch s Xm ¼ Figure 3-8 notes Equation To equal

Positive sequence closed open R1 =p3
ffiffiffi Interchange Ib and Ic (3-6) 2R1 I1
Negative sequence closed open R1 =p3
ffiffiffi As shown (3-5) 2R1 I2
HCB composite open closed R1 = 3 Interchange Ib and Ic (3-6) 2R1 I1 þ ðR1 þ 3R0 ÞI0
HCB-1 and SKB open closed 1:46R1 or As shown (3-5) 0:2I1 þ 0:462I2 þ
composites 0.191 ohms ðR1 þ 3R0 ÞI0
Data for Tap C of three taps available.

Eq. (3-5) to
VF ¼ ðR1 þ 3Xm ÞI1
pffiffiffi ð3-6Þ
þ ðR1  3Xm ÞI2 þ ðR1 þ 3R0 ÞI0
Note that the choice of design constant Xm ¼ R1 = 3
causes the I2 term in Eq. (3-6) to become 0. With
switch r closed and switch s open, the zero sequence
response of Eq. (3-5) and (3-6) is eliminated. The zero
sequence drop across R1 is 2=3R1 Iao  1=3R1
ðIb0 þ Ic0 Þ ¼ 0. The switches r and s are used in
Figure 3-8 as a convenience for description only.
Several typical sequence network combinations are
given in Table 3-1.

3.3 Sequence Voltage Networks

Sequence voltage networks may be constructed to

provide a single-phase output proportional to either Figure 3-9 Sequence voltage network.
positive or negative sequence voltage. A network in
common use is shown in Figure 3-9. Since this network
is connected phase to phase, there is no zero sequence
voltage effect.
The network is best explained through the phasor
diagram (Fig. 3-10). By design, the phase angle of Z þ
R is 608 lagging. For convenience, consider switches s
to be closed and switches r open. Impedance Z þ R is
thus connected across voltage Vab , and the autotrans-
former across voltage Vbc . With only positive sequence
voltages (Fig. 3-10a), the current Iab1 through Z þ R
lags Vab1 by 608. The drop Vby1 across the auto-
transformer to the tap is in phase with voltage Vbc
across the entire transformer. The tap is chosen so that
jVxb1 j ¼ jVby1 j. The filter output Vxy ¼ VF is the
phasor sum of these two voltages.
With only negative voltages applied (Fig. 3-10b), Figure 3-10 Phasor diagrams for the sequence voltage
Vxb2 is equal and opposite to Vby2 , that is, network of Figure 3-9 with ‘‘s’’ closed and ‘‘r’’ open.
50 Chapter 3

Vxy ¼ VF ¼ 0. Thus, this is a positive sequence net- recovers its nonconducting characteristics when the
work. reverse voltage falls below the zener value. They are
A negative sequence network can be made by used for surge protection, voltage-regulating functions,
reversing the b and c leads or, in Figure 3-9, by and other applications in which a distinct conduction
opening s and closing r. Then Figure 3-10a conditions level is desired.
apply to a negative sequence, giving an output VF ; Where conduction is desired in both directions with
Figure 3-10b conditions apply to a positive sequence a threshold at a level at which conduction occurs, the
with VF ¼ 0. back-to-back zener (Fig. 3-11c), commonly known as a
This interchange of b and c leads to either the volt trap or zener clipper, is used. The characteristics of
current or voltage networks offers a very convenient these devices are essentially the same in both the
technique for checking the networks. For example, the forward and reverse direction.
negative sequence current network should have no
output on a balanced power-system load but by 4.1.3 Varistor, Thermistor
interchanging the b and c leads it should produce full
output on test. The characteristics of the varistor are shown in Fig.
3-11d. It has a voltage-dependent nonlinear charac-
teristic. The thermistor depicted in Figure 3-11e is a
4 SOLID-STATE UNITS nonlinear device whose resistance varies with tempera-
4.1 Semiconductor Components
4.1.4 Transistor
Solid-state relays use various low-power components:
diodes, transistors, thyristors, and associated resistors In relaying, the transistor is used primarily as a switch.
and capacitors. These components have been designed For this function, it is shifted from a nonconducting to
into logic units used in many relays. Before these logic conducting state by the base current Ib. The transistor
units are described in detail, the semiconductor
components and their characteristics will be reviewed
(Fig. 3-11). Relays use silicon-type components almost
exclusively because of their stability over a wide
temperature range.

4.1.1 Diode
The diode (Fig. 3-11a) is a two-terminal device that
conducts in one direction but does not conduct in the
other. The device manifests a voltage drop for
conduction in the forward direction of approximately
0.7 V. The limit of voltage to be applied in the reverse
direction is defined by the rating of the diode. Failure
of the diode is expected if a voltage in excess of the
rating is applied in the reverse direction.
These devices are used in dc circuits to block
interaction between circuits, for ac test circuits to
generate a half-wave rectified current wave shape, or as
a protective device around a coil to minimize the
voltage associated with coil current interruption.

4.1.2 Zener Diode

The zener diode (Fig. 3-11b) differs from the diode
described above in having a sharp and reproducible
reverse breakdown voltage, called the zener voltage. If Figure 3-11 Semiconductor components and their charac-
the current is limited to within rated values, the diode teristics.
Basic Relay Units 51

Figure 3-13 Typical characteristic curves of transistor.

Figure 3-12 The transistor and equivalent electrical sym-

4.1.6 Unijunction Transistor
The unijunction transistor (Fig. 3-15) has two bases, b1
and b2, and one emitter e. When Ve reaches the peak
is nonconducting until Ib is increased to a value at value Vp, the device conducts and passes current Ie.
which the transistor conducts, and a collector current Current will continue to flow as long as Ve does not fall
Ic and emitter current Ie flow (Fig. 3-12). The emitter below the minimum value Vv. The unijunction
current Ie is the sum of Ib and Ic. Very small values of transistor is used for oscillator and timing circuits.
Ib are able to control much larger values of Ic and Ie
(Fig. 3-13).

4.1.5 Thyristor
The thyristor (Fig. 3-14) is a diode with a third
electrode (the gate). The thyristor is also known as a
silicon-controlled rectifier (SCR). With forward vol-
tage applied, the thyristor will not conduct until gate
current Ig is applied to trigger conduction. The higher
the gate current, the lower the anode-to-cathode
voltage (VF) required to start anode conduction. After
conduction is established and the gate current is
removed, the anode current IF continues to flow. The
minimum anode current required to sustain conduc-
tion is called the holding current IH. Figure 3-14 The thyristor and its characteristics.
52 Chapter 3

common use in the United States. In the commercial/

military system, the type of function is indicated by the
distinctive geometrical shape of the symbol. In solid-
state relaying, the name of the logic function is simply
written in a rectangle or block, or a distinctive symbol
such as ‘‘&’’ is used inside a block. The European
practice is similar to this. Convention dictates that
inputs are shown on the left-hand side and outputs on
the right-hand side. The symbols and terminology used
comply with IEEE Standard 91-1973 (ANSI Standard
U32 14-1973), ‘‘Graphic Symbols for Logic Dia-
Figure 3-15 The unijunction transistor and its character- grams.’’
istics. When a logic function has only two inputs, its
output is usually simple to determine. For three or
more inputs, particularly with combination logic
4.2 Solid-State Logic Units
functions, a logic or truth table offers a convenient
4.2.1 Basic Principles method of determining the output. A logic table for a
function with three inputs and one output is shown in
Solid-state logic units are combinations of solid-state
Figure 3-17. The table lists all possible combinations of
components designed to use dc voltage signals to
zeros and ones for the inputs. Each output could be 0
perform logic functions. A logic unit has only two
or 1, depending on the function.
states: no output, represented by 0 (zero), and output,
represented by 1 (one). Two logic conventions are used
to indicate the voltages associated with the 0 and 1 4.3 Principal Logic Units
states. In normal logic, 0 is equivalent to zero voltage
and 1 to normal voltage. In reverse logic, the In this section, the major units used in relaying will be
corresponding voltage equivalents are reversed; 0 is described. Detailed circuit descriptions will be kept to a
equivalent to normal voltage and 1 to zero voltage. minimum. For simplicity, the diagrams will show only
In positive logic, inputs and outputs are positive; in two inputs per function and include electromechanical
negative logic, both inputs and outputs are negative. contact equivalents.
Relay systems normally use positive logic, although
some elements may use negative signal inputs and 4.3.1 AND Unit
The AND logic element is shown in Figure 3-18. The
Logic units are shown diagrammatically in their
simplest type consists of forward-biased diodes and
quiescent state, that is, the normal or ‘‘at-rest’’ state. The
resistors (Fig. 3-18a). The symbolic representation and
quiescent state corresponds to the normally deenergized
electromechanical equivalents for this unit are given in
representation in electromechanical relay circuitry.
Figure 3-18b, the logic table in Figure 3-18c. The
forward-biased diodes shunt the output terminal, and
4.2.2 Logic Unit Representation
Logic units are represented by characteristic function
symbols (Fig. 3-16). Two sets of symbols are in

Figure 3-16 Examples of logic symbols. Figure 3-17 Example of logic table.
Basic Relay Units 53

Figure 3-18 AND logic.

Figure 3-20 OR logic.

no output voltage can appear unless all input diodes

have a reverse bias that equals or exceeds the forward
input from 0 to 1 or vice versa. For convenience, the
bias. Since inputs are either 0 or 1, there is no in-
symbol depicted in Figure 3-22a is replaced by that
between state that would allow partial output voltage.
shown in Figure 3-22b for negated inputs and that
Thus, the output is either 0 or 1, as shown in the logic
shown in Figure 3-22c for negated outputs.
table. Three variations of the AND element are
The NOT unit, together with its electromechanical
provided in Figure 3-19.
equivalent and logic table, is illustrated in Figure 3-23.
Although the NOT unit can be included in logic
4.3.2 OR Unit diagrams as a separate unit, it is usually combined with
The OR unit is shown in Figure 3-20. Again, the other units using the symbols shown in Figure 3-22.
simplest type of unit consists of resistors and diodes.
The symbolic representation and electromechanical
equivalents for the unit are illustrated in Figure 3-20b,
the logic table in Figure 3-20c. Since the diodes are not 4.3.4 Time-Delay Units
biased, an input voltage applied to any input will Time-delay units are used in the normal manner to
produce an output voltage at X. provide ON and/or OFF delays. The U.S. and
Three variations of the OR unit, comparable to those European symbolic representations are presented in
of the AND element, are shown together with their Figure 3-24. The X value in Figure 3-24 is the pickup
electromechanical equivalents in Figure 3-21a, b, and c. time, that is, the time that elapses between an input
By comparing Figure 3-21a with Figure 3-19b, it is clear signal being received and an output signal appearing.
that the inverse OR unit is equivalent to the negation The Y value is the dropout time, that is, the time after
AND. Similarly, the negation OR of Figure 3-21b the input signal is removed until the output signal goes
is equivalent to the inverse AND of Figure 3-19a. to 0. In Figure 3-24, W-X is the range of the pickup
time and Y-Z that of the dropout time. Any of the
4.3.3 NOT Unit values can be 0. Time values are always in milliseconds
unless otherwise indicated.
The negation, or NOT, unit (Fig. 3-22) is a frequently
used logic element. This unit changes the state of the

Figure 3-19 Variations of AND logic. Figure 3-21 Variations of OR logic.

54 Chapter 3

rent, phase angle, frequency, and so on) to determine if

any intolerable system conditions exist within the
relay’s zone of protection. The conventional functions
obtained by logic circuits are listed in Table 3-2.

5.1.1 Magnitude Comparison

There are two basic types of magnitude comparison
Figure 3-22 Negation symbol convention.
logic units: fixed-reference and variable-reference.

The logic circuit used for an instantaneous overcurrent
unit (Fig. 3-25) is basically a dc-level detector. Input
current from the current transformer secondary is
transformed to a current-derived voltage on the
Figure 3-23 NOT logic. secondary of the input transformer. This voltage is
limited by zener clipper Z1 and resistor R2. For low
input currents, the voltage is proportional to the
current, as determined by R1 and R3. The minimum
pickup is adjusted via the setting of R1. A low R1
setting diverts more current through R1 and R3, and
less to the phase splitter.
The phase splitter consists of a resistor-capacitor
network, transformer, and bridge rectifier. The output
voltage of the phase splitter is shown in the upper part
of Figure 3-25. When this voltage equals the zener
voltage of Z2, Z2 will conduct, providing a base
Figure 3-24 Examples of time delay units. current that turns on Q1. Q1 then turns on Q2,
providing an output current through D2 and R9. Q2
provides positive feedback through R7 and D1,
5 BASIC LOGIC CIRCUITS compounding the effect on the level detector.
In describing basic logic circuits, two types of diagrams
are used: the logic block diagram and logic circuit
schematic diagram. In the logic block diagram, the Table 3-2 Conventional Functions Obtained by Fault
Sensing and Data Processing Logic Circuits
units are represented by their logic symbols, and the
logic symbol blocks are interconnected to provide a Conventional Typical relay
complete functional representation of the system. In function Logic circuits types
the logic circuit schematic diagram, the elements are
Instantaneous Magnitude SI-T 50B
shown schematically. Unit interconnections are
overcurrent comparison with
depicted in the same way as in normal schematic fixed reference
diagrams. The logic block diagram is useful in showing Time overcurrent Magnitude 50D 51
the complete system in functional form; the logic comparison with
schematic circuit diagram indicates how the logic units fixed reference
operate. In the following discussion, logic schematic and time
diagrams will be used. Ground distance Magnitude SDG-T
comparison with
variable reference
5.1 Fault-Sensing Data Processing Units Phase distance Block-block SKD-T
Directional Coincident-time SRGU
In solid-state relays, fault-sensing and data-processing
(ring modulator)
logic circuits use power-system inputs (voltage, cur-
Basic Relay Units 55

Figure 3-25 Magnitude comparison dc level detector as an instantaneous overcurrent unit.

The dropout current can be adjusted by resistor R7, the restraint voltages is input to the desensitizer circuit.
normally set for a dropout/pickup ratio of about 0.97. When any combination of two restraint voltages is
Positive feedback provides the equivalent of snap smaller than the third restraint voltage, an output
action, and the 3% bandwidth prevents the equivalent produces a blocking action through D86, preventing
of chattering for current values close to minimum Q1 from turning on. When the operating voltage
pickup. This type of circuit could also be used for becomes larger than the largest restraint voltage,
overvoltage. reverse bias is applied to D86 through D38, turning
on Q1.
This logic unit discriminates between the value of an
5.1.2 Phase-Angle Comparison
operate voltage and the smallest of three restraint
voltages. Shown in Figure 3-26, this type of circuit Phase-angle comparator logic circuitry produces an
forms the decision logic element for the SDG ground output when the phase angle between two quantities is
distance relay described in Chapter 12. The restraint within certain critical limits. Either of these two
voltages (Vx, Vy, and Vz) and operating voltage are quantities, the polarizing (or reference quantity) and
connected in opposition through tunnel diode TD1 operating quantity, may be current or voltage.
and diodes D25, D26, and D27. When the operating Two types of phase-angle comparator logic circuitry
voltage is larger than any of the three restraint are in common use: block-block and coincident-time
voltages, current will flow through TD1. A small comparison.
current through TD1 drives it to a high voltage,
turning on Q1 and producing a voltage across the
Block-Block Comparison
output terminal. The tunnel diode characteristic
provides a sharp turn-on point, which serves as an The block-block type of phase-angle comparator uses
effective triggering action. the zero-crossing detector principle to generate square
Since double phase-to-ground faults may cause waves. Additional logic circuitry provides an output if
over-reach of the ground distance relay, a desensitizer the operating quantity leads the polarizing quantity.
circuit is included. This circuit consists of three Phase relations for the operating condition are shown
minimum voltage networks. A portion k of each of in Figure 3-27. An output is obtained if the operating
56 Chapter 3

Figure 3-26 Magnitude comparator circuit.

input leads the polarizing input by 0 to 1808. One half of the circuit of a block-block type of
Conversely, no output (restraint) occurs if the operat- comparator (Fig. 3-29) makes the comparison during
ing input lags the polarizing input by 0 to 1808. The the positive half-cycles. The input diodes, arrays DA
phase relation for this restraint condition is given in and DB, limit the input voltage to 1.5 V and the output
Figure 3-28. of transformers T1 and T2 to about 12 V.
Half-cycle square waves are generated at each zero For the operating condition shown in Figure 3-27,
crossing of the respective input quantities. The polarity the leading operating input makes the base of Q1
of the square waves is the same as that of the positive before the polarizing input can make the base
generating quantity during corresponding half-cycles. of Q3 positive. Thus, Q1 turns on first, which then

Figure 3-27 Phase relationship of block-block circuit for Figure 3-28 Phase relationship of block-block circuit for
operate condition. restraint condition.
Basic Relay Units 57

Coincident-Time Comparison (Ring Modulator)

Functioning like a biased bridge rectifier, the ring
modulator type of phase-angle comparator produces
an output when the operating quantity leads or lags the
polarizing quantity by 908 or less. This characteristic
makes the ring modulator applicable as a directional
sensing unit.
Figure 3-30 shows the operating principles of the
bridge under several input conditions. Current inputs
are depicted, but combinations of current and voltage
can also be used. Solid arrows indicate the input
operating quantities, open arrows the input polarizing
quantities. Actual current is the phasor sum of the
currents shown. The in-phase conditions are illustrated
in Figure 3-30.
In the bridge rectifier, two diodes are forward-
biased by the larger current, and the magnitude of the
output is determined by the smaller current. When the
operating current is larger (top half of Fig. 3-30a), D1
Figure 3-29 Block-block type phase-angle comparator and D3 are forward-biased, with the return through
circuit. R1 and R2 blocked by D4. The polarizing current is
shown in two parts, each one half of IPOL. The half
turns on Q2. Since Q5 has not been gated, it is in the going down through the transformer from point A
block state, permitting an output through Q2, R8, and flows backward through D3 and R1 to the polarizing
the output diodes. When Q2 turns on, D3 is reverse- terminal. However, since the operating current is
biased through D4 from the 20-V supply. This prevents larger, net current in D3 is forward. The output
the flow of base current from turning on Q4 as voltage IPOLR1 is proportional to the smaller current.
otherwise would occur as the lagging polarizing input If the operating current reverses and is still larger
becomes positive and turns on Q3. Since Q4 cannot than the polarizing current, D4 and D2 are forward-
turn on, a half-cycle of output occurs. Similarly, during biased, with the return through R2 and R1 blocked by
the negative half-cycle the leading operating input D1. Polarizing current going up from point A flows
provides an output in the other half of the circuit, backward through D2 and up through R2 (net current in
which connects through its negative half-cycle diode to D2 is forward). The output voltage IPOLR2 is reversed.
the output. With reversed but still smaller polarizing current,
If, however, the polarizing input leads the operating part of the polarizing current would flow up through
input (Fig. 3-28), the base of Q3 becomes positive R1, around through D1 (which is forward-biased by
before the base of Q1. Q3 turns on first and then turns the operating current) and back. The other part would
on Q4. The current flowing through Z1, Q4, R6, and flow up through R1, down through D3, and back.
R7 produces a voltage drop across R7. This voltage Again, the output voltage IPOLR1 is reversed.
drop gates thyristor Q5, causing it to conduct and If the polarizing current is larger, as in the bottom
short the output to negative. As the lagging operating half of Figure 3-30a, D1 is forward-biased through R1,
input becomes positive, it turns on Q1 and Q2 and and D4 is forward-biased through R2. If R1 equals R2,
(since Q5 is conducting) the current through Q2 and the net output from the polarizing current is 0. The
R8 is shunted to negative. The operating input that smaller operating current flows through D1, R1, R2,
remains when the polarizing half-cycle is completed and back through D4. Net current in D4 is forward
cannot produce an output, because Q5 continues to because of the larger value of polarizing current. The
conduct. Recovery is determined by the anode-to- net output then is IOP (R1 þ R2), or 2IOPR1. Reversing
cathode current, and R8 is set to allow sufficient either the polarizing or operating current will reverse
holding current from the 20-V supply to maintain Q5 the output voltage.
in a conducting state until the operating quantity is The output will not be 0 as long as the smaller
practically 0. current is above a threshold or pickup value. When the
58 Chapter 3

Figure 3-30 Principle of operation of ring-modulator type phase-angle comparator.

sum of the currents through a diode is 0, as through D3 components becoming larger, that is, the zero condi-
in Figure 3-30a, the output is still IOPR1. Any tendency tion no longer exists.
of either current to flow in another path, because D3 is Figure 3-30a shows the ring modulator operation
not conducting, will result in one of the two when the operating current is larger than the polarizing
Basic Relay Units 59

current and leads it by 908. At time 0, half of IPOL flows

up through R1, down through D3, and returns up the
lower half of the transformer to A. The other half flows
down through R2, up through D2, and down the upper
half of the transformer to A. The net output is 0 since
IOP is 0. As IOP increases from 0 to equal IPOL (point
P), IOP flows through D2, R2, R1, and D3, producing
an output of 2IOPR1, where R1 and R2 are equal.
When IOP equals IPOL, the current in D2 goes to 0. As
IOP becomes larger than IPOL, D1 conducts. The
negative IPOL all flows through R1; half passes through
D1, and the other half continues through D3. IOP flows
through D1 and D3, producing an output of
At time Q, when IOP again equals IPOL, the
polarizing current is about to become the larger
current and forward-bias D1 and D4. The output
changes to þ2IOPR1 and decreases. When IOP crosses
the zero axis, the output is 0. As the operating current
becomes negative, so does the output, which reaches a
maximum of 2IOPR. Figure 3-31 Breaker trip coil initiation circuit.
Further analysis shows that there is a maximum
positive or negative output each time IOP ¼ IPOL and
alternate one-half-cycle periods (4.17 msec) of positive Except for the transformer T2, the devices asso-
and negative outputs. These outputs are crosshatched ciated with Q4 provide security. Zener Z1 clips high-
in Figure 3-30b. Similar results are obtained if the voltage transients on the battery leads to a level of one-
polarizing quantity is greater than the operating third of the Q4 rating. This voltage clipping prevents
quantity, but with IOP leading IPOL by 908. the false operation of Q4 from surges and overvoltage.
The two-winding reactor LA-LB suppresses any
transients that could be transmitted through the
5.2 Amplification Units interwinding capacitance of T1 or between the trip
5.2.1 Breaker Trip Coil Initiator circuit and other logic circuit wiring. Zener Z4 prevents
shock excitation from setting up high-frequency
The breaker trip coil initiator circuit both provides oscillation, which might reverse the current through
power amplification for a trip coil and isolates the Q4 and return it to a blocking state.
control circuitry from the tripping energy source (the Capacitor C3 is initially charged through R9 and Z3
station battery). A typical circuit is shown in Figure 3-31. when the breaker or switch is closed, bypassing T2 to
Q1 turns on when the input voltage from the fault- avoid a false indication. When Q4 fires, C3 discharges
sensing and data-processing circuit exceeds 2 V. Q1 through Q4, Z2, and R8. This discharge provides a
then turns on Q2, allowing C2 to charge through R6. holding current for Q4 of about 1 msec, long enough
When the voltage across C2 reaches the ‘‘firing for the current through the inductive trip coil to reach
voltage’’ of the unijunction transistor Q3, the capacitor the required holding current for Q4.
energy discharges through T1. This discharge reduces
the voltage across the capacitor, turning off Q3 until
the charge on C2 builds up again. 5.3 Auxiliary Units
In this way, a repetitive train of pulses is generated
5.3.1 Annunciator Circuits
as long as the input signal exists. These pulses are
transformed through LA, Q4, T2 primary, LB, and Z4 Two types of circuits are used to provide light and
to trip the circuit breaker. The time delay of this circuit alarm indications: One is for circuit-breaker-trip
is approximately 1 msec. T1 has two secondaries, the operations and the other for general use.
second of which is connected to a similar Q4 circuitry Typical breaker-trip indicator and alarm logic are
for double trip. shown in Figure 3-32. Transformer T2 is in the trip
60 Chapter 3

Figure 3-32 Breaker trip indicator and alarm circuit.

circuit, as in Figure 3-31. The transformer core uses 5.3.2 Coordinating and Loop Logic Timers
square-hystersis loop material to produce a very small
Fixed time-delay timers are used extensively in logic
exciting current and negligible inductive reactance
circuitry. A typical circuit of this type is shown in
when saturated. When trip current flows (after Q4
Figure 3-34. With an input, Q1 is normally conducting
fires), the circuit of R1, C1, R2, and R3 stretches a 2-
and shorts C1 through R4. Removing the input turns
msec pulse at the secondary of T2 into 6 msec, at 20 V,
off Q1 and permits C1 to charge through R3 and R4.
at the output of Q2. The input signal turns on both Q1
When the voltage across C1 reaches the zener voltage
and Q2 to charge capacitor C2. When the voltage
of Z1 plus the potential hill of D1 and Q2, base current
builds up to the ‘‘intrinsic standoff ratio’’ of the
will flow, turning on Q2. Turning on Q2 removes
unijunction transistor (VP of Fig. 3-15), Q3 fires
voltage from the output. The fixed time interval is
and gates Q4, energizing the indicating light. The
between removal of input to removal of output.
conduction of Q4 also gates Q5 through R10 from the
Although normally used for short delays, the judicious
drop across R11. Q5 energizes the alarm relay. Even if
the indicating light circuit is open, Q5 will still be
A general indicator circuit is shown in Figure 3-33.
The normal condition is a 1 input, which makes Q1
conducting. For indication, the 1 is removed, turning
off Q1. Then C1 charges through R3 and R7. When
the voltage across C1 reaches the firing point of Q2, Q2
is turned on, gating Q4 and Q5 to energize the
indicating light and alarm relay.
The indicating lights are the solid-state equivalent of
mechanical indicating targets. Red lights are used to
indicate tripping or which sensing unit signaled a trip,
amber lights general alarms, blue lights testing. Sixty-
volt lamps operated at 48 V or 120-V lamps at 97 V
provide a filament life of more than 30,000 hr. Figure 3-33 Indicator and alarm circuit.
Basic Relay Units 61

negative. Q1 is not conducting. Its base, supplied

through R6, R4, and R2, is a negative. There is,
however, a clear output from the R3-R5-R8 voltage
Closing the set input switch S1 momentarily
reverses this condition. Voltage divider R1-R2 pro-
vides base drive to turn on Q1. The base drive for Q2 is
then shunted through Q1 to negative, and Q2 is turned
off. When S1 is opened, Q1 will remain on, through
voltage divider R2-R4-R6. Q2 will remain off, since
R5-R8 ties the base of Q2 to negative. Thus, with Q1
Figure 3-34 Typical logic timer circuit. on and Q2 off, there is a set output but no clear output.
When a momentary signal is applied to the reset
input, voltage divider R7-R8 provides base drive to
selection of values for R3, R4, C1, and Z1 provides a turn on Q2 again. This ties the base of Q1 to negative,
wide range of available time delays. Similar circuitry turning it off. Q2 then remains on, even after the reset
can provide a delay between an ON input and ON signal is removed. The unit is now back to its reset or
output, or other variations. Also, timers can be made normal state.
adjustable by making elements such as R3 adjustable. Figure 3-36a is a symbolic representation of a
normal flip-flop. A modification to the normal flip-flop
5.3.3 Toggle or Latching Circuits is to desensitize it by holding Q2 in a saturated
condition. When saturated, Q2 keeps conducting even
Toggle or latching circuits, known as flip-flops, are
when Q1 turns on. This prevents a spurious set signal
bistable units similar to a latched-in or toggle-type
from producing a set output. The modified flip-flop
relay. An operating signal will make the unit change
must first be ‘‘armed’’ by introducing an input arm
state; removal of the signal will leave the unit in the
signal (Fig. 3-36b). This signal removes the desensitiz-
new state. A momentary reset signal will restore the
ing bias from Q2 and allows it to turn off when the
unit to its original state. Normally, a momentary
normal set input signal is applied.
operating signal will change the output from 0 to 1 and
The flip-flop can also be combined with AND logic,
a momentary reset signal will change the output from 1 so that two or more separate set input signals must be
back to 0. The typical circuit shown in Figure 3-35 is
received simultaneously to produce an output. This
simplified to aid in the explanation of its operation.
modification may also be provided with desensitizing,
The circuit depicted is in the reset state with a
again requiring an arming signal. These modified flip-
‘‘clear’’ output and no ‘‘set’’ output. The voltage
flops are commonly used for the final trip logic unit in
dividers R3, R5, and R8 provide base voltage to Q2.
solid-state relaying systems.
Since Q2 is conducting, the set output is shunted to

Figure 3-35 Flip-flop circuit. Figure 3-36 Flip-flop logic symbols.

62 Chapter 3

5.3.4 Isolator and Buffer Circuits Q2, charging C2. The voltage across C2 triggers D1, as
described above. The pulses, rectified and filtered, are
Output and input isolators separate and electrically
applied to the base of Q3, turning it on and producing
isolate dc circuits between logic units. Used on the
an output. Zener Z1 provides surge protection clipping
input and output of each separately packaged relay,
at 20 V.
buffers protect the logic circuit from transients and
surge on interconnecting leads and circuitry. Both
isolator and buffer circuits protect solid-state relays Input Buffer
against undesirable operation on spurious signals. The input buffer circuit is shown in Figure 3-39. A
normal 20-V signal will result in approximately 90% of
Input Isolator the voltage appearing across R3 and capacitor C1.
A typical input isolator circuit is shown in Figure 3-37. When the voltage on C1 builds up to around 5 to 7 V,
A 20-V input to the pulsing circuit of R1-R3-C1-D2 current flows through Z2, D1, and R4. Q1 then turns
charges the capacitor C1. When the capacitor voltage on, which produces an output. R5 is required when the
reaches the breakdown voltage of the four-layer diode output drives a PNP stage, but is omitted for an NPN
D2, a pulse is transmitted through T1. The discharge stage. For internal logic circuitry, Q1 can be turned on
of C1 turns off D2 until the voltage across C1 builds by an unbuffered input.
up again. Thus, a series of pulses continues as long as There are three types of buffering: (1) A high-
the input signal exists. Zener Z1 provides surge frequency, high-voltage surge on the input, such as the
protection clipping at 20 V. The pulses are rectified 1.0- to 1.5-MHz, 2500-V standard test surge, is
and accumulated on C2. C2-R4-R5 provide a steady dc dropped across R1 and clipped to 20 V by Z1; (2) all
input to Q1 until the input is removed. Q1 conducts, signals of 150 to 200 msec are delayed by means of R1-
turning on Q2, providing a 20-V output. R2-C1; (3) a minimum threshold voltage of 6 V is
required to turn on Q1. Thus the maximum ‘‘0 level’’
Output Isolator voltage is 6 V. A bona fide signal must exist for at least
150 msec. This buffer is described further in Chapter 4
The input section of the output isolator circuit (see Fig. 4-15).
(Figure 3-38) is similar to the breaker-trip coil
initiation circuit shown in Figure 3-31. The output
Output Buffer
isolator circuit differs in that a four-layer diode D1,
rather than a unijunction transistor, provides a pulse The output buffer circuit is shown in Figure 3-40. An
chain through T1. An input voltage turns on Q1 and input greater than 2 V turns on Q1 and Q2 to provide

Figure 3-37 Input isolator circuit.

Basic Relay Units 63

Figure 3-38 Output isolator circuit.

approximately 18 V output. C1 provides a 75-msec 6 INTEGRATED CIRCUITS

delay through the unit. High-voltage, high-frequency
transients on the output are limited and clipped by R6 The next trend in solid-state relaying was toward the
and the 24-V zener Z1. Should the output be shorted, use of linear and digital integrated circuits to replace
Q2 is protected by the current-limiting action of R6. the discrete transistor circuits described previously. An
overview of the linear integrated circuit operational
amplifier and its application to basic relay units
Optical Isolator
The isolation of solid-state circuits and components
from input and output signals is also accomplished 6.1 Operational Amplifier
with the use of optocoupler devices. This integrated
circuit component uses an internal light-emitting diode Figure 3-41 shows the equivalent circuit of a basic
(LED) and a photon detector to transmit signals, operational amplifier. The triangle symbol is used for
providing optical isolation between inputs and out- this device. The supply voltages +Vcc (generally +15
puts. Vdc) with a ‘‘common’’ of 0 V are not shown. The

Figure 3-39 Input buffer circuit. Figure 3-40 Output buffer circuit.
64 Chapter 3

Figure 3-43 An operational amplifier with negative feed-

Figure 3-41 The equivalent circuit of an operational back.

If Zf is much smaller than Ri , the input resistance, the

assumption is that no current flows in the a or b
input terminals are a and b; b is the noninverting input terminals, so that
since a positive voltage produces a positive output. A
positive voltage on a, the inverting terminal, will yield Iin ¼ If ð3-9Þ
a negative output. The drops around the Zf feedback loop are
The output e0 is amplified by the open-loop gain A
so that  eref þ en þ If Zf þ e0 ¼ 0
e0 ¼ Aen ¼ Aðeb  ea Þ ð3-7Þ eref  en  e0 ð3-10Þ
If ¼
The plot in Figure 3-42 shows that a small differential
change drives the amplifier into saturation since the Equating Eqs. (3-8) and (3-10) with the assumption of
open-loop gain A is very large. Eq. (3-9) provides
Most applications use negative feedback. In    
Zf Zin
Figure 3-43, where Zf is connected from the output en ¼ eref  ein  e0 ð3-11Þ
to the inverting input a, Iin can be determined from the Zf þ Zin Zf þ Zin
drops around the input loop,
Substituting Eq. (3-7) and solving for e0 , we obtain
 ein þ Iin Zin  en þ eref ¼ 0    
1 Zin Zf
ein þ en  eref ð3-8Þ e0 þ ¼ eref  ein ð3-12Þ
Iin ¼ A Zin þ Zf Zf þ Zin
If we assume that A is very large, 1/A approaches 0
and is much less than
Zin þ Zf

Zf Zf
e0 ¼ 1 þ eref  ein ð3-13Þ
Zin Zin

Equation (3-13) is the general operational amplifier

equation with negative feedback. If it is substituted
into Eq. (3-11), the solution for en will equal 0. Thus, a
and b terminals are of the same relative potential. The
Figure 3-42 Saturation characteristics for circuit in Figure inverting input terminal (a) is referred to as virtual
3-41. ground.
Basic Relay Units 65

6.2 Basic Operational Amplifier Units

A number of basic units are derived from a single

operational amplifier to use in relay circuits. These are
described without the additional components required
for accuracy, stability, or compensation.

6.2.1 Inverting Amplifiers

The inverting amplifier of Figure 3-44 is the circuit of
Figure 3-43 with terminal b connected directly to
common (0 V). From Eq. (3-13), we get
e0 ¼  ein ð3-14Þ
If resistors are used as shown in Figure 3-44, the
output e0 is the opposite of the input modified by the
scale factor

Figure 3-45 A noninverting amplifier and voltage follower

6.2.2 Noninverting Amplifiers unit.
If the a input is reduced to 0 through Rin (Fig. 3-45a)
and the input applied to terminal b instead of eref , Eq.
(3-13) reduces to Another version is the voltage follower shown in
  Figure 3-45b. Rf approaches 0, a short circuit, and Rin
Rf approaches infinity, an open circuit. The gain factor
e0 ¼ 1 þ ein ð3-15Þ
The input and output are in phase with a scale factor Rin
approaches 0 so that the scale factor
1þ Rf
Rin 1þ
If Rf is made very large compared to Rin , then for a
approaches unity. Thus, the output voltage e0 equals
sine wave input, the output essentially will be an
or follows ein . In this circuit, the input impedance seen
inphase square wave to provide a squaring circuit.
by ein essentially is infinite and no current flows into
the b terminal.

6.2.3 Adders
An adder unit (Fig. 3-46) has two separate inputs
through Ra1 and Ra2 to the negative terminal a with
terminal b at 0. Equation (3-13) reduces to
Rf Rf
e0 ¼  ea1  ea2 ð3-16Þ
Ra1 Ra2
If Ra1 ¼ Ra2 ¼ Rf , then the output equals the negative
Figure 3-44 An inverting amplifier unit. of ea1 þ ea2 .
66 Chapter 3

Figure 3-48 An integrator and low-pass filter unit.

Figure 3-46 An adder unit.

6.2.4 Subtractors and since

The basic circuit is shown in Figure 3-47. The voltage ein
at the plus terminal of the operational amplifier will be if ¼ ¼ iin
Z ð3-19Þ
eb e0 ¼  ein dt
Rf þ Rin Rin C
this circuit is a simple low-pass filter. Considering
Substituting this in Equation (3-13), we obtain
magnitudes only,
Rf 1
e0 ¼ ðeb  ea Þ ð3-17Þ Zf ¼
Rin 2pfC
If Rf ¼ Rin , then e0 ¼ eb  ea . so that Eq. (3-14) becomes
je0 j ¼  jein j ð3-20Þ
6.2.5 Integrator and Simple Low-Pass Filter Thus, as frequency increases, the magnitude of e0
With a capacitor as the feedback component, the decreases.
inverting amplifier of Figure 3-44 becomes an inte-
grator (Fig. 3-48): 6.2.6 Differentiator and Simple High-Pass Filter
Z Unit
e0 ¼  if dt ð3-18Þ This circuit is shown in Figure 3-49 and is the inverted
amplifier circuit with a capacitor in the input circuit
iin ¼ if ¼ C ð3-21Þ
so that
e0 ¼ Rf C ð3-22Þ
with magnitudes from Eq. (3-14), where
Zin ¼ and Z f ¼ Rf :

Figure 3-47 A subtractor unit. je0 j ¼ 2pfCRf jein j ð3-23Þ

Basic Relay Units 67

Figure 3-49 A differentiator and high-pass filter unit.

This is a simple high-pass filter since as f decreases, je0 j


6.2.7 Phase-Shift Units

A variety of phase-shift units are obtained using
capacitor and variable resistor combinations. These
are illustrated in Figure 3-50. A phase-angle range of
90 to 1808 is obtained with Zf adjustable from 0 to
908 (Fig. 3-50a), 1808 to 2708 with Zin adjustable
from 08 to 908 (Fig. 3-50b). Inverting operational
amplifiers are used in both of these circuits.
Noninverting amplifiers with the RC network
connected as a voltage divider (Figs. 3-50c and d)
provide a phase-angle range of 0 to þ908 or 0 to 908,
depending on the position of R and C.
Figure 3-50 Phase shift units.
6.2.8 Level Detectors
Figure 3-51a shows a level detector using the opera-
tional amplifier in the differential mode. From Eq.
(3-7), we have R3. This voltage is higher than with just R1 and R2: e0
approaches 0 when ein exceeds the eref . This causes eref
e0 ¼ Aðeref  ein Þ ð3-24Þ
to be lowered to a potential determined by the divider
and relationship of R1 in series with the parallel combina-
  tion of R2 and R3. Thus, the voltage at which e0
R2 switches from high to low is greater than that when it
eref ¼ Vcc ð3-25Þ
R1 þ R2 switches from low to high. This is illustrated in the
example in the lower half of Figure 3-51b.
As illustrated in Figure 3-42, a change in the level of
ein ðea Þ slightly before or below eref ðeb Þ will cause the
amplifier to go into either negative or positive
saturations, respectively. With eref formed by the R1-
6.2.9 Active Filters
R2 voltage divider, e0 becomes low with ein above eref ,
and high when ein is less than eref . A typical active filter unit is shown in Figure 3-52.
Hysteresis is obtained with positive feedback High ‘‘Q’’ circuits, different gains, and resonant
through resistor R3 (see Fig. 3-51b). With e0 large, frequencies are easy to obtain. Inductance is not used
eref is determined by the voltage divider consisting of in these circuits. The filters can be cascaded and are
R2 in series with the parallel combination of R1 and unaffected by loading.
68 Chapter 3

6.3.1 Instantaneous Overcurrent Unit

An operational amplifier instantaneous overcurrent
unit is shown in Figure 3-53. Input current i is
converted to a proportional voltage through shunt R
and filtered by an active bandpass filter (OA1). The
pickup at other than the system frequency is signifi-
cantly higher to minimize harmonic effects. OA2 is an
adjustable gain inverting amplifier with a gain of

K ¼

The amplified signal Ki is rectified by OA3 and OA4.

When the signal from OA2 is negative, OA3 forces its
output positive, while the input to the (þ) terminal of
OA4 will be negative through R7. This back-biases
diode D1 to disconnect the OA3 output to OA4. OA4
acts as a voltage follower with its output negative and
following the (þ) terminal input. When OA2 output
goes positive, the output of OA3 goes negative and
applies a negative input to the (þ) terminal of OA4
through D1. With R5 equal to R6, OA3 is a unity gain
inverter with the input to OA4 negative when the OA3
input is positive.
Figure 3-51 Level detector units.

6.3 Relay Applications of Operational Amplifier

Three protective relay applications illustrate the use of

the basic operational amplifier units described. The
relaying inputs from current and voltage transformers
are converted to low-level signals by shunts or
auxiliary transformers.

Figure 3-52 A multiple-feedback band-pass filter unit. Figure 3-53 An instantaneous overcurrent unit.
Basic Relay Units 69

6.3.2 Sequence Networks

Sequence networks can be designed by using opera-
tional amplifiers. A negative sequence circuit is shown
in Figure 3-54. From Chapter 2, we have
I2 ¼ ðIa þ a2 Ib þ aIc Þ
With phase-shift units (Fig. 3-50), Ib is shifted 2408 and
Ic 1208. With adder units, the final output is
e0 ¼ ðIa þ Ib ff240 þ Ic ff120 Þ
and with R ¼ 1 O,
Figure 3-54 An operational amplifier negative sequence
1 network.
e2 ¼ e0 ¼ ðIa þ a2 Ib þ a1c Þ
Positive sequence and composite filters can be designed
following the same techniques. The top circuit X has an adjustable noninverting
amplifier and a level detector. If the sine wave is of
sufficient magnitude to exceed the level detector setting
(R4 and R5), the level detector output switches to 0
6.3.3 Threshold Squarers and Square-Wave
during the positive half-cycle as shown in the wave
traces. P1 determines the magnitude at which the
These basic units are used in phase-comparison pilot output switches. At low currents, the output remains 1,
systems. A typical circuit is shown in Figure 3-55. The whereas at high currents, the output is a square wave.
outputs provide square waves at a low level for keying The middle circuit Y is similar to X except that an
to a remote terminal, and at a high level for local inverting amplifier is used to provide a positive output
comparison. from a negative input current. At high currents, the

Figure 3-55 A threshold squarer and square-wave detector.

70 Chapter 3

level detector switches from 1 to 0 during the negative level detector threshold, causing output D to become 1.
half-cycle input. When the square wave becomes 1 during the opposite
The lower circuit Z generates a symmetrical square half-cycle, the charge rate of R11, C, is high enough to
wave at low currents. At no current, it has a 1 output. prevent further switching of the level detector.
The circuit is similar to the X circuit. P3 is much larger
than R3, and R8 much larger than R9 to provide a
high-gain and low-level detector-switching voltage. 7 MICROPROCESSOR ARCHITECTURE
Square-wave detection is accomplished with the
operational amplifier timing circuit whose output is Relay design has evolved from electromechanical and
D. Resistor R11 should be greater than R10 so that the solid-state to microprocessor. The functions of electro-
capacitor C discharge rate is less than the charge rate. mechanical sensing units, sequence networks, and
With no square wave (i.e., no current), then capacitor solid-state logic units are performed through the
C will remain fully charged, causing the level detector processing of digital signals. The microprocessor
following it to remain switched in the low state. component, integrated with RAM and ROM devices,
However, if a square wave exists, then the zero and software programs make up the basic unit in
transitions allow capacitor C to discharge below the microprocessor relay design.
Protection Against Transients and Surges

1 INTRODUCTION The magnitude of the coupled voltage VL is CM/

(CM þ CG) per unit of Vn, as long as RL and RS are
The sporadic damped phenomena that occur in very high. Vn is the effective noise voltage and RR the
electrical systems are generally described as transients effective load resistance of the noisy lead. The lower
and surges. In this book, the two terms are considered RL and RS are, the lower the transient voltage. If RL
synonymous and will be used interchangeably. In some and RS are so low that their effect predominates, the
references, however, transients refer to those phenom- voltage on the signal lead becomes approximately
ena related to lumped system parameters; surges refer RTCM (dVn / dt), where RT is the parallel equivalent of
to those phenomena related to distributed parameters. RL and RS, and dVn / dt the rate of change of the noise
For any disturbance in an electrical circuit, such as the voltage. The voltage on the signal lead cannot however
opening or closing of a switch or breaker, the exceed CM/(CM þ CG) per unit, regardless of the rate
associated damped transients may be either oscillatory of change of the noise voltage.
or unidirectional. Surges also appear as traveling In some systems such as those used in solid-state
waves with a distinct propagation velocity. In such relaying, where negative—rather than ground—is the
cases, wave reflections may produce voltages substan- ‘‘common,’’ the equivalent circuit is that shown in
tially greater than the forcing voltage that initiated the Figure 4-2a. The basic circuit is rearranged in Figure
phenomenon. Lightning surges must be considered as 4-2b.
well. With rare exception, however, experience indi-
cates that only high-voltage systems need be protected
against lightning.
From a relaying standpoint, the effect of transients
and surges on secondary control circuits is of principal
importance. Primary transients affect secondary cir-
cuits through common electrical connections, such as
‘‘ground’’ circuits and electrostatic or electromagnetic
induction, as well as current and voltage transformers.

1.1 Electrostatic Induction

A simplified version of electrostatic pickup is shown in

Figure 4-1. An error signal is introduced into the Figure 4-1 Equivalent circuit for electrostatic induction
‘‘signal lead’’ via the mutual coupling capacitance CM. with common ground return.

72 Chapter 4

total induced loop voltage is MdI/dt, where M is the

effective mutual impedance between the two circuits
and dI/dt the rate of change of current I. Transposing
the signal circuit will reduce the induced voltage, as
shown in Figure 4-4.

1.3 Differential- and Common-Mode


Surges can be classified into two modes: differential

(also known as normal or transverse) and common
(also known as longitudinal).
Differential-mode surges produce voltage on a pair
of conductors in the same way as a legitimate signal.
Differential-mode signals are illustrated in Figures 4-1,
4-2, 4-3, 4-8, and 4-9.
Common-mode surges produce equal voltages on a
pair of conductors, with respect to some common
references. Common-mode surges are generated as
Figure 4-2 Equivalent circuit for electrostatic induction shown in Figure 4-7. Common-mode voltage is also
without common ground return. produced by the circuit shown in Figure 4-2 if CM1
equals CM2 and CG1 equals CG2.
Differential-mode surges are more likely to produce
1.2 Electromagnetic Induction misoperation of equipment, whereas common-mode
surges are more likely to produce dielectric failure.
Figure 4-3 illustrates electromagnetic pickup. Flux- (Note also that purely common-mode surges, when
linking of the signal pair, resulting from current flow in applied to unbalanced circuits, will produce a differ-
an adjacent circuit, induces a false signal voltage. The ential-mode component and vice versa.)

Figure 4-4 Transposing the signal circuit to minimize

Figure 4-3 Electromagnetic induction. electromagnetic induction.
Protection Against Transients and Surges 73


2.1 Capacitor Switching

Primary circuit transients are frequently generated by

capacitor switching and are substantially more severe
when interruption is accompanied by restriking.

2.1.1 Single-Bank Capacitor Switching

Figure 4-5 demonstrates what happens when a
capacitor bank is energized by closing a switch. A
high-frequency, high-magnitude current I flows. The
capacitance of the bus and connected apparatus causes
the same phenomenon to occur when a bus section is
energized. Unless precautions are taken to avoid
transients, such switching can cause 5- to 6-kV peaks
in secondary circuits. Figure 4-6 illustrates what
happens if restriking occurs when a capacitive current
Figure 4-6 Transients generated by opening a capacitive
is interrupted. circuit.
At the instant of interruption (current zero), full
voltage VC is trapped on the capacitor bank. This
voltage cannot change unless further current flows. process of equalization, considerable overshoot occurs,
The source voltage VB, however, continues to vary and both VB and VC approach three times normal line-
sinusoidally. If the interruption cannot support the to-neutral peak voltage. The current flow, immediately
recovery voltage, a restrike occurs and current flows following the restrike, is also very high. The current
again. (The recovery voltage is VC  VB.) The most oscillates at the natural frequency of the circuit and
unfavorable instant of restriking is shown in Figure 4-6. decays with time, as governed by the circuit time
With continuity reestablished, VC equals VB. In the constant.

2.1.2 Back-to-Back Capacitor Switching

Back-to-back capacitor switching consists of the
energization of one bank of capacitors adjacent to a
previously energized bank. Back-to-back capacitor
switching is much like the energization of a single
bank, except that the effective inductance is generally
very much lower. The capacitance, on the other hand,
is only somewhat lower, since it is the series combina-
tion of the bank capacitance and capacitance of the
unit or units that were energized before the switch was
closed. For these reasons, the magnitude and fre-
quency of the current are generally much higher for
back-to-back energization than single-bank energiza-

2.2 Bus Deenergization

Figure 4-5 Transients generated by energizing a capacitive Bus dropping is similar to capacitor bank deenergiza-
circuit. tion, except the capacitance C is very much smaller.
74 Chapter 4

Current magnitude is also generally smaller, and the

frequency is higher. When a simple disconnect is used
to drop the bus, the nonlinearity and prolonged
existence of the restriking arc cause significant
electrical noise. These characteristics together with
the large voltages and currents that accompany
restrikes produce one of the most severe surge
influences in a substation. Surges of up to 8 kV have
been measured in secondary circuits during disconnect

2.3 Transmission Line Switching

Transmission line switching is also similar to capacitor

bank switching, except for the distributed nature of the
inductance and capacitance of the line. The inrush
current tends to be substantially less than that for
capacitor bank switching. Frequency is inversely
proportional to the length of the transmission line.

Figure 4-7 Surge in secondary leads during disconnect

switch restriking on a capacitance voltage transformer.
2.4 Coupling Capacitor Voltage Transformer
(CCVT) Switching

These transformers contain capacitance voltage-divid- 2.5 Other Transient Sources

ing networks. After energization, deenergization, and
restriking, they are subjected to the same high- Many other switching-type operations generate tran-
frequency, high-current phenomenon experienced in sients: unequal pole-closing of a circuit breaker, fault
the other cases of lumped capacitance switching. Even occurrence, fault clearing, load-tap changing, line
in a well-designed capacitance voltage transformer, reactor deenergization, series capacitor gap flashing
there is perceptible capacitance between the high- and reinsertion, and so forth. In general, the peak
voltage and low-voltage windings (Fig. 4-7). At the magnitude of such transients is substantially less than
high frequencies associated with capacitor device for the phenomenon described above.
switching, the impedance of this capacitance will be
small. A surge voltage is developed during disconnect
restriking around the path g-g0 -p-q-x or y and is
roughly equivalent to L di/dt  M di/dt + Ri. (L and
R are the inductance and resistance of the ground lead
of the voltage device, and M is the mutual impedance 3.1 Direct Current Coil Interruption
between the ground lead and voltage leads.) If M
equals L, the total surge voltage reduces to Ri. In During interruption of an inductive circuit, such as a
practice, M can never equal L, but it will approach it if relay coil, the L di/dt effect may produce a large
the potential leads are placed as close as possible to voltage across the coil (Fig. 4-8). In general, the
the ground lead. This arrangement will lessen the voltage will be greatest at the instant of interruption.
transient voltage between the voltage leads and Voltage magnitude will generally be independent of the
ground. supply circuit characteristics and equal to the differ-
Since voltage transformers are inductive devices, ence between the extinction voltage of the interrupting
they are not subject to this phenomenon. contact and the battery voltage. The surge voltage
Protection Against Transients and Surges 75

Figure 4-9 Transients produced in adjacent circuits by dc

circuit energization.

burden. The surge repeats during each transition from

saturation in one direction to saturation in the other.
The voltage appearing at the secondary consists high-
magnitude (possibly several kV) spikes with alternating
Figure 4-8 Transients produced by interruption of an polarity that persist for a few milliseconds eve half-
induction circuit. cycle.

increases as a function of the speed with which the

3.4 Grounding of Battery Circuit
interruptor forces current zero. Although voltages in
excess of 10 kV have been generated across 125-V coils
When a ground occurs on the dc system, the
in laboratory tests. 2.5 kV is a more typical value.
distributed and lumped capacitance of a system may
cause sensitive devices to operate. Figures 4-10 and
3.2 Direct Current Circuit Energization 4-11 illustrate trip circuit behavior in the event of an
accidental ground. Comparable phenomena can cause
Energizing a circuit that is capacitively coupled to sensitive close circuits and tripping relays to malfunc-
adjacent or nearby circuits can produce a transient in tion.
the latter circuits (Fig. 4-9). When switch 1 is closed,
VR appears as a false signal across the effective
resistance of the adjacent circuit. Initially, full battery 4 PROTECTIVE MEASURES
voltage appears across the coupled circuit. This voltage
the decays exponentially, in accordance with the RC 4.1 Separation
tim constant. 4.1.1 Physical Separation
Noise in critical circuits can be controlled effectively by
3.3 Current Transformer Saturation physically separating quiet and noisy circuits. Since
mutual capacitance and mutual inductance are inverse
Current transformer saturation, which may produce logarithmic functions of distance, small increases in
ver high secondary voltage, is caused by high primary distance produce substantial decreases in circuit
current, poor current transformer quality, or excessive interaction.
76 Chapter 4

Similarly, control circuits should be routed perpen-

dicular to noisy circuits. For example, a cable duct
should be run perpendicular to a high-voltage bus
when possible. Another way of effectively controlling
surges is to group circuits with comparable sensitiv-
ities. Low-energy-level circuits, especially, should be
grouped together and placed as far as possible from
power circuits.

4.1.2 Electrical Separation

Circuits can, of course, also be separated electrically.
For example, surges can be controlled by the
discriminate application of inductance to block con-
duction of high-frequency transients into protected
regions. This principle is illustrated by the filter circuit
shown in Figure 4-12. High-frequency transients are
diverted harmlessly to ground.
Transformer isolation (Fig. 4-13) puts an effective
common-mode barrier between segments of a system.
Figure 4-10 Accidental ground on battery positive. High capacitance from each winding to ground and
low capacitance from winding to winding further
reduce common-mode interaction between windings.

Figure 4-12 Choke coil isolation.

Figure 4-13 Transformer isolation for common mode

Figure 4-11 Accidental ground on trip lead. voltage.
Protection Against Transients and Surges 77

4.2 Suppression at the Source secondary. Silicon carbide devices can be used in this
protective function.
4.2.1 Resistor Switching
Transient voltages can be kept comparatively low by 4.2.3 Suppression by Termination
equipping disconnects and circuit breakers with
resistors that are inserted during operation of the Figures 4-1, 4-2, 4-3, and 4-9 illustrate the value of
device. For reasons of economy, this arrangement is reduced input impedance RL in restricting the magni-
occasionally used to restrict the surge level in substa- tude and/or duration of transients. However, if RL is
tions. reduced, the energy requirement for operation is
increased, and more heat is generated when legitimate
inputs are applied.
A small capacitor offers another method of redu-
4.2.2 Parallel Clamp
cing input impedance at high frequency, with little
The surge associated with coil interruption can be effect at 50 or 60 Hz or on dc. This device neither
virtually eliminated by paralleling the coil with a zener requires a higher input energy for operation nor
diode. Where an extended dropout time is undesirable, generates heat. One such widely used capacitor is a
a varistor may be substituted for the zener diode 0.01-mf ceramic capacitor. It limits a 2500-V, 1-MHz
arrangement. Although the varistor allows a higher surge, with a 150-O source to 350 V peak to peak.
surge than the zener diode, its limiting action is Short leads to the capacitor are imperative.
satisfactory. When sensitive relays, trip circuits, and close circuits
The zener diode Z1 in Figure 4-14 performs a dual exist in a substation, the capacitance on the dc must be
surge function. First, it minimizes the inductive ‘‘kick’’ restricted if false operations are to be avoided when a
produced by the deenergization of the auxiliary coil. ground occurs.
Also, when one of the contacts (Trip A in Figure 4-14)
closes, the voltage induced in TC BKR A resulting
from the interruption of current flow when 52a A 4.3 Suppression by Shielding
opens cannot cause the auxiliary relay to pick up
undesirably. Current path I will have no detrimental A signal lead that is shielded and has one or more
effect. Zener Z1 will allow forward voltage of only grounds will have increased capacitance to ground CG
approximately 0.7 V, which is insufficient to operate (Fig. 4-1). For high RS and RL values, this increase in
the auxiliary relay. This scheme prevents undesirable capacitance to ground reduces the ‘‘false’’ signal
tripping of breaker B. voltage VL that results from the presence of an
The transient associated with extreme ac saturation adjacent noisy lead. If a shield were used in Figure
of a current transformer can also be squelched by 4-2, it would surround the signal lead and the common
introducing a voltage-limiting device across the negative and would be grounded in one or more
locations. This arrangement tends to force CM1 =CG1 to
equal CM2 =CG2 , and any capacitively induced signal
voltage across RL to be 0.
Grounding a shield at both ends allows shield
current to flow. Shield current resulting from magnetic
induction will tend to cancel the flux that created it.
The net effect of the shield on the signal lead is to
reduce the noise level. Both ends should not be
grounded if the shield is the signal return path.

4.4 Suppression by Twisting

Measures that cause the signal and return leads to

occupy essentially the same space minimize the effect
of differential-mode coupling (Fig. 4-4). As shown by
the polarity marks, twisting a pair of leads cancels the
Figure 4-14 Zener Z1 applied for surge suppression. effect of adjacent circuit flux. Also, twisting the signal
78 Chapter 4

lead and negative causes CM1 to equal CM2 , and CG1 to

equal CG2 (Fig. 4-2). This technique substantially
reduces the influence of the adjacent noisy lead.
A combination of shielding and twisting effectively
minimizes the influence of surges in adjacent circuits.
For circuits properly treated with SPP capacitors at the
terminal blocks, shielding is not required for static
relaying circuits inside a panel or switchboard. These
SPP capacitors are 750-V dc oil-filled capacitors.
Shielded twisted pair conductors are required for
low-energy-level circuits routed outside a panel.
One lead of the shielded twisted pair is normally the
signal lead. The other lead (except where it is sensing Figure 4-15 A standard input buffer circuit for solid-state
contact status) connects the negatives of the two relays.
devices. Within a panel, electrostatic coupling is the
only significant intercircuit transient influence. A single
ground on the shield, therefore, is sufficient. For
consistency, ground should be at the input end. (open circuit) first peak, which decays to 1250 V in 6 or
more msec. The buffer can also withstand a sustained 7-
Vdc input, or a high dc input voltage of sufficient
4.5 Radial Routing of Control Cables duration to produce a minimum 4000-msec-V product
(for example, 20 V for 200 ms).
Circuits routed into the switchyard from the control Buffering low-energy-level circuits greatly decreases
house should not be looped from one piece of the susceptibility of static relays to surge damage or
switchyard apparatus to another with the return malfunction and, in general, eliminates the need for
conductor in another cable. Rather, all supply and shielding circuits inside a relaying panel.
return conductors should be in a common cable. This
arrangement avoids the large EMI (electromagnetic
induction) associated with the large flux loop that
would otherwise be produced. 4.7 Optical Isolators

Optical isolators can provide excellent electrical

4.6 Buffers separation between two circuits. Figure 4-16 describes
an input isolator. When the LED is conducting, as a
Another effective method of delaying and desensitizing result of a voltage being applied at the input, base drive
a circuit is to use a buffer (Fig. 4-15). Without causing is provided for the phototransistor and it conducts,
the transistor to conduct or damaging any element, supplying a ‘‘logical’’ input to the protected equip-
this buffer can accommodate a test source operating at ment.
1 to 1.5 MHz with a 150-O source impedance directly Similar isolation is accomplished for an output by a
across the input (differential mode) and a 2500-V circuit such as that of Figure 4-17.

Figure 4-16 Optically isolated input.

Protection Against Transients and Surges 79

Figure 4-17 Optically isolated output.

4.8 Increased Energy Requirement breaker or operate any other devices. An auxiliary relay
designed to pick up at 71 V or more will not respond to
Surges can also be endured by raising the threshold a single ground on a dc circuit with a maximum
voltage or energy level at which operation occurs. The operating voltage of 140 V, regardless of the magnitude
equivalent circuits of Figures 4-10 and 4-11 show that of the capacitances on the system. (Note that the
half-maximum battery voltage, applied through the higher-voltage design will not solve the problem shown
appropriate capacitance, must not be allowed to trip the in Figure 4-9; the higher-energy-level restriction will.)
Instrument Transformers for Relaying

1 INTRODUCTION load current. Whenpdelta-connected

ffiffiffi current transfor-
mers are used, the 3 factor must be considered.
Instrument transformers are used both to protect Although the performance required of current
personnel and apparatus from high voltage and to transformers varies with the relay application, high-
allow reasonable insulation levels and current-carrying quality transformers should always be used. The
capacity in relays, meters, and instruments. Instrument better-quality transformers reduce application pro-
transformer performance is critical in protective blems, present fewer hazards, and generally provide
relaying, since the relays can only be as accurate as better relaying. The quality of the current transformers
the information supplied them by the instrument is most critical for differential schemes, where the
transformers. Standard instrument transformers and performance of all the transformers must match. In
relays are normally rated at 5 or 1 A; 100, 110, or these schemes, relay performance is a function of the
120 V; and 50 or 60 Hz. accuracy of reproduction—not only at load currents,
Where the relays operate only on current or voltage but also at all fault currents as well.
magnitude, the relative direction of current flow in the Some differences in performance can be accommo-
transformer windings is not important. Relative dated in the relays. In general, the performance of
direction (and, therefore, polarity) must be known, current transformers is not so critical for transmission
however, where the relays compare the sum or line protection. The current transformers should
difference of two currents or the interactions of several reproduce reasonably faithfully for faults near the
currents or voltages. The polarity is usually marked on remote terminal, or at the balance point for coordina-
the instrument transformer but can be determined if tion or measurement.

2.1 Saturation
For large-magnitude, close-in faults, the current
One major criterion for selecting a current transformer transformer may saturate; however, the magnitude of
ratio is the continuous current ratings of the connected fault current is not critical to many relays. For
equipment (relays, auxiliary current transformers, example, an induction overcurrent relay may be
instruments, etc.) and of the secondary winding of operating on the flat part of the curve for a large-
the current transformer itself. In practice, with load magnitude, close-in fault. Here it is relatively unim-
current normally flowing through the phase relays or portant whether the current transformer current is
devices, the ratio is selected so that the secondary accurate, since the timing is essentially identical. The
output is around 5 A (or 1 A) at maximum primary same is true for instantaneous or distance-type relaying

82 Chapter 5

for a heavy internal fault well inside the cut-off or

balance point. In all cases, however, the current
transformer should provide sufficient current to
operate the relay positively.

2.2 Effect of dc Component

The presence of dc in the primary current can be

particularly detrimental to ct performance. This
phenomenon is described in Section 6, ‘‘Direct Current
Saturation.’’ Having a dc component in fault current
on an ac power system is a decaying phenomenon. If a
ct is going to saturate due to the dc component of fault
current, it will do so in the first few cycles. Until this
effect takes place, the fidelity of transformation is
reasonably good, and instantaneous overcurrent and
distance relays may perform their task before the ct
performance collapses.
Following the disappearance of the dc component
of fault current, the behavior of the ct will again Figure 5-1 The equivalent circuit and phasor diagram of a
improve. The error of transformation may then be current transformer.
predicted by using one of the methods described in
Section 4.

purposes is usually made on the basis of sinusoidal

3 EQUIVALENT CIRCUIT fundamental quantities. Although this approach is
highly simplified, the equivalent diagram is an excellent
An approximate equivalent circuit for a current tool for picturing the phenomenon and estimating the
transformer is shown in Figure 5-1. Current is stepped approximate performance to be expected.
down in magnitude through the perfect (no-loss)
transformation provided by windings ab and cd. The
primary leakage impedance (ZH) is modified by n2 to
refer it to the secondary. The secondary impedance is
ZL; Rm and Xm represent the core loss and exciting 4 ESTIMATION OF CURRENT TRANSFORMER
components. PERFORMANCE
This generalized circuit can be reduced, as shown in
Figure 5-1b. ZH can be neglected, since it influences A current transformer’s performance is measured by
neither the perfectly transformed current IH/n nor the its ability to reproduce the primary current in terms of
voltage across Xm. The current through Xm, the the secondary; in particular, by the highest secondary
magnetizing branch, is Ie, the exciting current. The Rm voltage the transformer can produce without satura-
branch produces a negligible influence. tion and, consequently, large errors. Current trans-
The phasor diagram, with exaggerated voltage former performance with symmetrical (no dc) primary
drops, is shown in Figure 5-1c. In general, ZL is current can be estimated by
resistive and ZB is resistive or has a lagging angle. Ie
lags Vcd by 908 and is the prime source of error. Note Formula
that the net effect of Ie is to cause IL to lead and be The current transformer excitation curves
smaller than the perfectly transformed current IH/n. The ANSI transformer relaying accuracy classes
Any simple equivalent diagram for a current
transformer is, at best, crude. Exciting current is The first two methods provide accurate data for
accompanied by harmonics that, in turn, produce analysis; the latter gives only a qualitative appraisal.
harmonic relay currents. An analysis for application All three methods require determining the secondary
Instrument Transformers for Relaying 83

voltage Vcd that must be generated voltage VS of 100(2.0 þ 0.31), or 231 V. Equation (5-2),
solved for Bmax, will determine whether the current
Vcd ¼ VS ¼ IL ðZL þ Zlead þ ZB Þ ð5-1Þ
transformer can reproduce this current
Bmax ¼ ¼ 70;000 lines=in:2
VS = rms symmetrical secondary induced voltage 4:4466063:16400
Fig. 5-1)
Therefore, the current transformer should have iron
IL = maximum secondary current in amperes that will not saturate below 70,000 lines/in.2. Since the
(symmetrical) current transformer in this example uses high-perme-
ZB = connected burden impedance in ohms ability silicon steel, it will not saturate with symme-
ZL = secondary winding impedance in ohms trical primary current.
Zlead = connecting lead impedance in ohms

4.2 Excitation Curve Method

4.1 Formula Method A typical excitation curve for a current transformer is

shown in Figure 5-2. These data represent rms currents
The formula method uses the fundamental transformer obtained by applying rms voltage to the current
equation transformer secondary, with the primary open-
circuited. The curve gives the approximate exciting
VS ¼ 4:44 fANBmax 108 ðvoltsÞ ð5-2Þ current requirements for a given secondary voltage.
f = frequency in hertz
A = cross-sectional area of the iron core in
square inches
N = number of turns
Bmax = flux density in lines per square inch

Both the cross-sectional area of the iron and its

saturation density are sometimes difficult to obtain.
Current transformers generally use silicon steels, which
saturate from 77,500 to 125,000 lines/in.2. The lower
figure is typical for current transformers built before
1947; a value of 100,000 is typical of most transformers.
The formula method consists of determining VS
using Eq. (5-1), then calculating Bmax using Eq. (5-2).
If Bmax exceeds the saturation density, there will be
appreciable error in the secondary current.
Assume, for example, that a 2000:5, high-perme-
ability silicon steel transformer has 3.1 in.2 of iron and
a secondary winding resistance of 0.31 O. The max-
imum current for which the current transformer must
operate is 40,000 A at 60 Hz. The relay burden,
including the secondary leads, is 2.0 O. Will this
current transformer saturate?
If the current transformer does not saturate, the
secondary current IS would be 40,000 divided by 400,
or 100 A, since N equals 400. Thus, the current Figure 5-2 Exitation curves for a multiratio bushing current
transformer should be able to produce a secondary transformer with an ANSI accuracy classification of C100.
84 Chapter 5

In theory, when making the phasor addition, the angle

of the burden and the exciting branches should be
taken into account. This refinement is not necessary,
however, since it is obvious from the curve of Figure
5-2 whether or not the current transformer would be
operating in the saturated region.
An alternative approach would be to use a ratio of
60:3, or 20 (the 100:5 tap), with the higher burden of
the 3-A relay tap. If we use this ratio, N equals 20, and
IS equals 3 A to operate this relay:

VS ¼ IL Z total ¼ 363 ¼ 9 V

From the excitation curve (Fig. 5-2), Ie equals 0.5 A,

Figure 5-3 Excitation curve method. and NIe is 10. The primary pickup current IH would be

IH ¼ 60 þ 10 ¼ 70:0 A
With this method, a curve relating primary current
to secondary current can be developed for the tap, lead This value is closer, but still too high.
length, and burden being used (Fig. 5-3). Any value of Now suppose that the breaker has two sets of
primary current can then be entered on the curve to current transformers, with the secondaries connected
determine the expected value of secondary current. in series. Then each current transformer carries one-
The following examples will illustrate some of the half the burden, or 1.54 O on the 3-A tap. This value is
problems encountered in estimating current transfor- slightly more than one-half of 3.0 O because of the
mer performance using the excitation curve method. secondary resistance of the added transformer. Using
the 100:5 tap, we obtain

Example 1 Phase Relays N ¼ 20 turns

The breaker has a multiratio 600:5 bushing current IL ¼ 3 A
transformer and the feeder is protected with over- VS ¼ 361:54 ¼ 4:62 V per transformer
current relays. The relays should operate for approxi-
mately 60 A rms symmetrical primary current. The
Then, from Figure 5-2, we have
total burden on the current transformer, including the
current transformer secondary resistance, is 1.6 O per
Ie ¼ 0:33
phase when the relays are on the 6-A tap, and 3 O per
phase on the 3-A tap. The excitation curve for the NIe ¼ 6:6
transformer is shown in Figure 5-2. IH ¼ 3620 þ 6:6 ¼ 66:6 A
One approach would be to use a current transfor-
mer ratio of 60:6, or 10 (the 50;5 tap), to take Although this alternative offers some improvement, IH
advantage of the lower burden on relay tap 6: is not as close to the desired 60 A as might have been
expected. In both cases, the current transformer is
N ¼ 10 turns
operating on the straight-line part of the characteristic,
IL ¼ 6 A to operate the relay making significant improvement difficult. On the other
VS ¼ IL Z total ¼ 661:6 ¼ 9:6 V hand, two 50:5 current transformers in series would
show a marked improvement over the 50:5 ratio. Here
From the excitation curve for VS of 9.6 V, Ie would IH, calculated by the above methods, is 71 A. While
be 6 A, and NIe equals 60. Therefore, the primary much better than 120 A, this value is still not as good
pickup current is as the 66.6-A pickup obtained using the two current
IH ¼ NIS þ NIe ¼ 60 þ 60 ¼ 120 A transformers with a 100:5 ratio.
Similar evaluations can be made for other config-
This value is considerably higher than the 60 A desired. urations.
Instrument Transformers for Relaying 85

Figure 5-4 Connections for Example 2 illustrating calcula-

tion of current transformer performance.

Example 2 Phase and Ground Relays Figure 5-5 Equivalent circuits and distribution of currents
for a ground fault with the connections of Figure 5-4 in
The following example, shown in Figure 5-4, will
Example 2.
determine the minimum primary current that will
operate the phase and ground relays.
Thus, the phase-a relay circuit must supply 0.6 plus
PHASE RELAYS. For the phase relays, the total phase 0.6 plus 0.5, or 1.7 A. Given the phase relay impedance
burden Z equals 0.68 plus 0.08, or 0.76 O, where 0.08 is of 0.68 O and current transformer impedance of 0:08 O,
the current transformer secondary resistance on the or 0:76 O total, the phase-a current transformer must
100:5 tap (N ¼ 20): supply
IL ¼ 5 A(to operate the relay on the 5-A tap) VS ¼ 11 þ ð1:760:76Þ ¼ 12:3 V
VS ¼ 0:7665 ¼ 3:8 V Ie ¼ 0:8 A (from Fig. 5-2)
From Figure 5-2, we get IL ¼ 1:7 þ 0:8 ¼ 2:5 A
IH ¼ 2:5620 ¼ 50 A primary
Ie ¼ 0:28 A
IH ¼ NðIL þ Ie Þ ¼ 20ð5 þ 0:28Þ ¼ 105:6 A primary Thus, 50 A is required to operate the ground relay. If
the exciting requirements of the three current trans-
If we neglect the exciting current (Ie), this value formers had been ignored, the current required to
would become 20 times 5, or 100 A primary, when operate the ground relay would have been estimated to
using the 100:5 current transformer ratio. be 0.5 times 20, or 10 A primary. From this, it is
apparent that such may be a significant factor.
GROUND RELAYS. If we assume the ground current Using the 200:5 tap on the current transformer
flows only in phase a, the equivalent circuit is shown in could improve sensitivity here. Dramatic improvement
Figure 5-5. To obtain 0.5 O through the ground relay, would also be possible if a modern, low-impedance
with its impedance assumed here of 22 O, 11 V must be ground relay were substituted.
produced across the ground relay. If we neglect the
small unknown voltage across the phase relays, this
ground relay voltage will appear across the phase-b 4.3 ANSI Standard: Current Transformer
and phase-c current transformers to excite them from Accuracy Classes
the secondary side. From Figure 5-2, an Ie of 0.6 A
develops 11 V across these current transformers. The The ANSI relaying accuracy class (ANSI C57-13) is
accuracy required, generally, does not warrant described by two symbols—letter designation and
correction for the small phase relay drop. However, voltage rating—that define the capability of the
such a correction could be made on a trial-and-error transformer.
basis. The letter designation code is as follows:
86 Chapter 5

Relaying accuracy classification.

Mechanical and thermal short-time (1-sec) ratings.
Both ratings define rms values that the transfor-
mer is capable of withstanding. For mechanical
short-time ratings, the rms value is that of the ac
component of a completely displaced primary
current wave. The thermal 1-sec rating is the rms
value of the primary current that the transformer
will withstand with the secondary winding short-
circuited, without exceeding the limiting tem-
perature of 250 8C for 55 8C-rise transformers, or
350 8C for 80 8C-rise transformers. The short-
time thermal current rating for any period of 1 to
5 sec is determined by dividing the 1-sec current
rating by the square root of the required number
of seconds.
Resistance of the secondary winding between the
Figure 5-6 ANSI accuracy standard chart for class C winding terminals. Data should be presented in a
current transformers. form that allows the value for each published
ratio to be determined.
For T-class transformers, the manufacturer should
C: The transformer ratio can be calculated. supply typical overcurrent ratio curves on
T: The transformer ratio must be determined by rectangular coordinate paper. The plot should
test. be between primary and secondary current, over
the range from 1 to 22 times normal current, for
The C classification covers bushing current transfor-
all standard burdens up to the one that causes a
mers with uniformly distributed windings, and any
ratio error of 50% (Fig. 5-7).
other transformers whose core leakage flux has a
negligible effect on the ratio within the defined limits.
The T classification covers most wound-type trans-
formers and any others whose core leakage flux affects
the ratio appreciably.
The secondary terminal voltage rating (Fig. 5-6) is
the voltage that the transformer will deliver to a
standard burden at 20 times normal secondary current,
without exceeding a 10% ratio error.
Figure 5-6 shows the secondary voltage capability
for various C-class current transformers, plotted
against secondary current. With a transformer in
the C100 accuracy class, for example, the transformer
ratio can be calculated, and the ratio error will not
exceed 10% between 1 and 20 times normal secondary
current if the burden does not exceed 1:0 O
ð1:0 O65 A620 ¼ 100 VÞ.
ANSI accuracy class ratings apply only to the full
winding. When there is a tapped secondary, a
proportionately lower-voltage rating exists on the taps.

4.3.1 Current Transformer Data

The following current transformer data, required for
relaying service application, should be supplied by the Figure 5-7 Typical overcurrent ratio curves for a T class
manufacturer: current transformer.
Instrument Transformers for Relaying 87

For C-class transformers, the manufacturer should so, what corrective action can be taken to reduce the
also supply typical excitation curves on log-log error to 10% or less?
coordinate paper. The plot should show excita- The current transformer secondary winding resis-
tion current and secondary terminal voltage for tance may be ignored because the C200 relaying
each published ratio from 1% of the accuracy accuracy class designation indicates that the current
class secondary terminal voltage to a voltage (not transformer can support 200 V plus the voltage drop
to exceed 1600 V) that will cause an excitation caused by ct internal secondary resistance at 20 times
current of five times normal secondary current rated current, for a 50% power-factor burden. The ct
(Fig. 5-2). secondary voltage drop may be ignored then if the
secondary current does not exceed 100 A:
The ANSI standard burden is defined with a 50%
power factor. These standard ohmic burden values are 800
N¼ ¼ 160
identified in Figure 5-6. When fewer than the total 5
number of turns are in use on the C-class current 12; 000
IL ¼ ¼ 75 A
transformer, only a portion of that burden can be 160
supplied without exceeding the 10% error. Maximum
The permissible burden is given by (from Eq. 5-3)
permissible burden is defined mathematically by
NP Vcl
NP Vcl ZB ¼
ZB ¼ ð5-3Þ 100
100 800 (proportion of total
NP ¼ ¼ 0:667
where 1200 turns in use)
ZB ¼ permissible burden on the current transformer
NP ¼ turns in use divided by total turns 0:667ð200Þ
Vcl ¼ current transformer voltage class ZB ¼ ¼ 1:334 O
Standard relaying burdens are listed in Table 5-1. Since the circuit burden, 2:4 O, is greater than the
B-0.1, B-0.2, B-0.5, B-0.9, and B-1.8 are standard calculated permissible burden, 1:334 O, the error will
metering burdens. These burdens are defined with a 0.9 exceed 10% at the maximum fault current level (75 A).
power factor. Consequently, it is necessary to reduce the burden, use
The following example shows current transformer a higher current transformer ratio, or use a current
calculations using ANSI classifications: the maximum transformer with a higher relaying accuracy class
calculated fault current for a particular line is 12,000 A. voltage.
The current transformer is rated at 1200:5 and is to be
used on the 800:5 tap. Its relaying accuracy class is
C200 (full-rated winding); secondary resistance is
0:2 O. The total secondary circuit burden is 2:4 O at a
60% power factor. Excluding the effects of residual 5 EUROPEAN PRACTICE
magnetism and dc offset, will the error exceed 10%? If
In Europe, current transformers are described in terms
of protection and measurement classes. The protection
Table 5-1 Standard Relay Burden Designations classes are those of most interest in relaying, and they
carry the designation P, a maximum error of 5 or 10%,
Characteristics for 60-Hz and 5-A secondary circuit a corresponding volt-ampere burden, a rated current,
Standard and an accuracy limit factor. For example, a 30-VA
burden Impedance class, 5P10, 5-A ct is compatible with a 30-VA
designation (r) VA Power factor continuous burden at 5 A. This corresponds to a 6-V
output. It produces no more than 5% error at
B-1 1.0 25 0.5 1066 ¼ 60 V. The permissible burden is 30=
B-2 2.0 50 0.5 ð565Þ ¼ 1:2 O.
B-4 4.0 100 0.5
Three types of ct’s are defined by the TPX, TPY,
B-8 8.0 200 0.5
and TPZ designations.
88 Chapter 5

5.1 TPX

The TPX is a nongapped core current transformer with

a 0.5% ratio error and secondary time constant of 5 sec
or more. It may be used with other TPX or TPY ct’s in
all types of protection applications.

5.2 TPY

This ct has a gapped core and secondary time constant

of 0 to 10 sec. It has a ratio error of +1% and larger
cost than the TPX. Its transformation of the dc
component of fault current is not as accurate as the
TPX. It may be combined with other TPX or TPY ct’s
in any relaying application. Its advantage is that its
remanent flux is quite small compared to that of a
nongapped core ct.

5.3 TPZ

TPZ ct’s have a linear core with a secondary time

constant of 60+6 msec for 50-Hz and 50+5 msec for
60-Hz applications. This provides a very short dc
collapse time, making the ct suitable for breaker failure
applications in which the overcurrent supervision is
susceptible to dc influence. When used in combina-
Figure 5-8 Current immediately after fault inception.
tions, it should be used only with other TPZ ct’s. It has
a +1% ratio error at the rated primary current.
result of dc saturation. Note the improvement in
performance as the dc diminishes.
6 DIRECT CURRENT SATURATION If VK  6:28 IRT, the dc component of a fault
current will not produce current transformer satura-
To this point, current transformer performance has tion. In this expression,
been discussed in terms of steady-state behavior only,
without considering the dc component of the fault VK ¼ voltage at the knee of the saturation curve,
current. Actually, the dc component has far more determined by extending the straight-line por-
influence in producing severe saturation than the ac tions of the curve to find their intersection
fault current. The dc component arises because (1) the I ¼ symmetrical secondary current in amperes rms
current in an inductance cannot change instanta-
neously and (2) the steady-state current, before and
after a change, must lag (or lead) the voltage by the
proper power-factor angle.
Figure 5-8 shows the current immediately following
fault inception for two cases: fully offset and with no
offset. In the fully offset case, the fault is assumed to
occur at the instant that produces the maximum dc
component. In the second case, the fault occurs at a
time that produces no dc offset.
Figure 5-9 shows an example of the distortion and
reduction in the secondary current that occurs as a Figure 5-9 Direct current saturation of current transformer.
Instrument Transformers for Relaying 89

R ¼ total secondary resistance in ohms

T ¼ dc time constant of the primary circuit in cycles
Here, T ¼ (LP/RP)f, where
LP ¼ primary circuit inductance in henries
RP ¼ primary circuit resistance in ohms
f ¼ frequency
Direct current saturation is particularly significant
in bus differential relaying systems, where highly
differing currents flow to an external fault through
the current transformers of the various circuits.
Dissimilar saturation in any differential scheme will
produce operating current.
Figure 5-10 shows how current transformer satura-
tion relates to time. Severe current transformer
saturation will occur if the primary circuit dc time
constant is sufficiently long and the dc component
sufficiently high. Curves d, e, and f of Figure 5-10 show
that the dc component requires substantially greater
flux than that needed to satisfy the ac component.
Time is required to reach saturation flux density.
This time can be estimated from Figure 5-11, as
From the current transformer excitation curve for
the tap in use, determine VK from the intersec-
tion formed by extending the two straight-line
segments of the curve. Note that both axes must
have the same logarithmic scales, as is illustrated
in Figure 5-2.
Calculate VK/IRT.
Obtain t/T from Figure 5-11.
Calculate t, the time to saturate.
VK must be modified if residual flux is to be
considered. For example, with a residual flux of 90%,
the saturation voltage value must be multiplied by 1
minus 0.9, or 0.1, to determine the earliest time to
saturation. This will give a conservative value for time
to saturation.


Any iron-core device will retain a flux level even after

the exciting current falls to 0. Superimposed on this
Figure 5-10 Current transformer flux during assymetrical
residual flux are variations in core flux, dictated by the fault (one cycle time constant).
current transformer secondary current and secondary
Figure 5-12 shows the importance of previous defined by point a. If a symmetrical sinusoidal primary
loading history on current transformer residual flux current starts to flow, requiring a flux variation as
level. Suppose a current transformer has a residual flux shown, the pattern between a and A will be traced out.
90 Chapter 5

Figure 5-12 Residual flux in current transformer.

The average value of the dc exciting current with this

pattern is Ia . This current flows in the secondary and
has no counterpart in the primary. It decays with the
time constant associated with the secondary circuit. At
the completion of this transient, the pattern has moved
to cC with an equal flux variation that is symmetrical
around the vertical axis. The pattern continues to be
traced out. If the circuit were now interrupted, the
residual flux would have the value existing at the
moment of interruption, which is quite different from
the initial value assumed. In fact, any value of flux
between 0 and the saturation level may be retained in
the core, depending on previous events.
Reducing the residual flux to 0 requires the
application of a secondary voltage high enough to
produce saturation, followed by a gradual reduction of
the voltage to 0.
A current transformer with an air gap in the core
has a fairly low residual flux, approximately 10% of
saturation density. Residual flux for current transfor-
mers with no intentional air gap is approximately 90%
of saturation density (max).
Air-gap current transformers do not saturate as
rapidly as devices without air gaps subjected to equal
current and burden. However, the magnetizing current
is higher, resulting in greater ratio and phase-angle
After interruption of the primary current, residual
flux decays very slowly (taking approximately 1 sec),
and the secondary current collapses slowly. Also, air-
Figure 5-11 (a) Current transformer time to saturate. (b) gap units are more costly to manufacture, since the
Expanded scale of (a). small air gap must be both accurate and maintainable.
Instrument Transformers for Relaying 91

Although in theory residual flux can cause relaying

problems, there have been very few documented cases
in which the residual flux has caused a relay


The MOCT (magneto-optic current transducer)

relieves many of the problems associated with current
transformation. This device utilizes the Faraday effect
to produce a high-accuracy analog output that is not
influenced by iron saturation. The Faraday effect is the Figure 5-14 Faraday effect current sensing.
rotation of the plane of polarization when plane-
polarized light is sent through glass in a direction
parallel to an applied magnetic field. This is illustrated
in Figure 5-13. The angle of rotation is directly
proportional to the strength of the magnetic field. 9 VOLTAGE TRANSFORMERS AND
In the application of this principle to current COUPLING CAPACITANCE VOLTAGE
measurement, the transmission line current is the TRANSFORMERS
source of the magnetic field (see Fig. 5-14).
The strength of the field is directly proportional to Voltage transformers (formerly called potential trans-
the instantaneous current magnitude. By placing the formers) and coupling capacitance voltage transfor-
rotator (Faraday-effect sensor) in proximity to mers are selected according to two criteria: the system
the transmission line conductor and comparing the voltage level and basic impulse insulation level
angle of rotation of a light beam, a voltage is generated required by the system on which they are to be used.
that is directly proportional to current. Under ANSI, two nominal secondary voltages, 115
This voltage is then used as an input to protective and 120 V, are allowed for voltage transformers;
relays at a level and in a way that is virtually identical pffiffiffi the
pffiffiffi line-to-neutral values are 115= 3 and
to that which is used when current transformers supply 120= 3. The applicable voltage depends on the
the relay through a current-to-voltage transformation. primary voltage rating, as given in ANSI C57.13.
The nominal secondary voltages for coupling capaci-
tance voltage transformers are 115 and 66.4 V.
Most protective relays applied in the United States
have standard voltage ratings of 120 or 69 V, depend-
ing on whether they are to be connected line to line or
line to neutral.

9.1 Equivalent Circuit of a Voltage Transformer

The equivalent circuit of a voltage transformer (vt) is

shown in Figure 5-15. Since regulation is critical to
accuracy, the circuit may be reduced to that shown in
Figure 5-15b. The phasor diagram of Figure 5-15c has
greatly exaggerated voltage drops to emphasize that,
for typical transformers and burdens, the secondary
voltage usually lags the ‘‘perfectly transformed’’
primary voltage and is deficient in magnitude. Typical
rated maximum errors for these devices are 0.3, 0.6,
Figure 5-13 Faraday rotator concept. and 1.2%. Voltage transformers have excellent tran-
92 Chapter 5

Figure 5-17 Simplified schematic of a coupling capacitor

voltage transformer.

Figures 5-17 and 5-18 illustrate the source of the

subsidence transient in the ccvt. In Figure 5-18,
elements L and C generally contain stored energy
when a disturbance, such as a fault, occurs on the
Figure 5-15 The equivalent circuit and phasor diagram of a primary. Because of the ‘‘ringing’’ tendency inherent in
voltage transformer. the RLC circuit, a sudden short circuit on the primary
does not produce an instantaneous collapse of the
voltage applied to the relays. The extent and duration
sient performance, faithfully reproducing abrupt of the deviation from the perfectly transformed voltage
changes in the primary voltage. depend on the values of R, L, and C. Other transients
are introduced by the presence of ferroresonant
suppression circuits and the relays themselves.
9.2 Coupling Capacitor Voltage Transformers A voltage transformer is not significantly affected
by comparable transients and will reproduce primary
Coupling capacitor voltage transformers (ccvt’s) and transients with excellent fidelity. Modern ccvt’s, such
bushing capacitor voltage transformers are less expen- as the PCA-7, have capabilities approaching those of
sive than voltage transformers at the higher voltage the voltage transformer.
ratings, but may be inferior in transient performance. The subsidence transient of the ccvt may influence
With these voltage devices, a subsidence transient the behavior of some relays. Solid-state phase and
accompanies a sudden reduction of voltage on the ground distance relays, used in a zone 1 direct trip
primary. This voltage may be oscillatory at 60 Hz or function, may be seriously affected by the temporary
some other frequency, or it may be unidirectional. A excessive reduction of voltage during the decay period.
representative severe secondary transient is shown in These relays either must be designed with a special
Figure 5-16. provision that allows the subsidence transient to be
ignored, be time-delayed to override the transient
period or they must have their reach shortened
sufficiently to avoid false tripping.

Figure 5-16 A typical subsidence transient of older type Figure 5-18 Equivalent diagram of a coupling capacitor
coupling capacitor voltage transformers. voltage transformer.
Instrument Transformers for Relaying 93

Table 5-2 ABB ccvt’s The impedance of capacitance voltage transformers

Capacitance Transient
should not be high enough to produce erroneous
Steady- p.u. response behavior in the static compensator distance relays.
state Max. (1 p.u. ¼ (% voltage) Excessive impedance may cause false tripping for a
Cost accuracy burden 0.006 at reverse fault. For this reason, bushing voltage devices
Type p.u. (%) (VA) 115 kV) 8 msec 16 msec rated below 230 kV should not be used with solid-state
PCA-5 — 1.2 200 1.0 23.5 12 distance relays. Bushing voltage devices are, in general,
PCA-7 1.57 1.2 200 4.1 7 6.2 seldom used. Their burden capability is limited and
PCA-8 1.0 1.2 200 1.0 23.5 12 transient performance poor.
PCA-9 1.94 0.3 400 4.1 15 10.5
PCA-10 1.26 1.2 200 4.1 17.5 11
PCM-X 2.51 0.3 400 4.1 12 9.5 9.3 MOVT/EOVT

Voltage sensing also may be accomplished using fiber-

Table 5-2 describes a group of ABB ccvt’s with
optic technology. The MOVT uses the Faraday effect
differing costs and performance. The transient
described above, sensing the current flowing through a
response column indicates, in general, the degree of
capacitor stack connected from line to ground.
compatibility with different relaying systems. High-
Another voltage-sensing device, the EOVT, uses a
speed, direct trip, restricted reach relays require either
Pockel cell rather than the Faraday rotator. Its
the use of the fast response ccvt’s or provision in the
principle uses light from an optical fiber, which is
relaying system to accommodate or eliminate the effect
passed through a special crystal that produces equal
of the error. The transient response data are based on
components in the X and Y directions. An electric field
the percentage of voltage remaining at the ccvt output
causes one of these components to be retarded, and
terminals at the times indicated following sudden
this results in a phase difference between the two
reduction of primary voltage from rated voltage to 0,
components. This, in turn, changes the light intensity
with initiation at zero crossing. The burden is as
at the sensor fiber in proportion to the electric field.
defined in ANSI C93.1.
With additional refinements, it produces an analog
output proportional to the electric field present, and
this is, in turn, proportional to the instantaneous
magnitude of the voltage at the point of measurement.


Neutral ‘‘inversion,’’ in which ground becomes exter-

nal to the system voltage triangle, can occur on
ungrounded systems with a single potential transfor-
mer connected line to ground. Figure 5-19 shows the
possible voltage across an unloaded voltage transfor-
mer. Note that an Xc/Xm ratio of 3 would theoretically
cause an infinite voltage across the voltage transfor-
mer. Such a situation never occurs, of course, because
Xm reduces as saturation occurs.
By loading the transformer carefully, this large,
sustained overvoltage phenomenon can be avoided.
Caution should always be exercised when the second-
Figure 5-19 Neutral inversion on an ungrounded power ary voltage of the transformer is used for synchronism
system. check, since the loading will cause a phase shift.
Microprocessor Relaying Fundamentals

1 INTRODUCTION and devise ways for the microprocessor to accomplish

the tasks in the right order and to cause comparisons
From the power-system viewpoint, microprocessor or to be made based on the correct voltages and currents
numerical relays are not unlike electromechanical, without the error associated with data skew. Data
solid-state, or digital relays. Currents and voltages skew is introduced by the comparison of quantities
must be measured and compared with set points, or taken nonsimultaneously.
with each other, and action must be deferred or To allow the sampling of a fixed quantity rather
initiated. Other inputs such as received carrier, 52b than a rapidly changing quantity, the sample-and-hold
switch position, choice of pilot system, etc. temper the (S/H) circuit is usually used in numerical relays. An
action. Figure 6-1 shows a simplified block diagram of example of this is shown in Figure 6-2. Since even the
a typical microprocessor relay. simplest relaying function requires that multiple inputs
Numerical relays must work within the framework be read, a multiplexer is used. This is a device that
of data sampled moment by moment. Currents, for allows all the input quantities to be sampled (read) one
example, are not treated on a continuous basis, but at a time.
they, like all of the input quantities, are sampled one at The microprocessor requires that the information
a time at as fast a speed as the data handling and be presented to it in digital form, usually an 8- or 16-bit
storage hardware can accommodate. word. The conversion process from the analog signal
The microprocessor has afforded us in protective (which is simply a scaled dc quantity that is repre-
relaying the remarkable capability of sampling vol- sentative of the sampled quantity) to the digital signal
tages and currents at very high speed, manipulating the is accomplished with an A/D converter. Many varieties
data to accomplish a distance or overcurrent measure- of these devices have been used over the years. The
ment, retaining fault information and performing self- range and sampling rate required dictate the choice for
checking functions. The utilization of this new a particular design for a protective relay.
technology has also presented us with new challenges The microprocessor accepts the sampled data and
regarding the manner in which information is handled stores it for future use in RAM (random access
and manipulated. memory). The data are acted on by algorithms or
With multiple electromechanical or solid-state comparisons defined by the program memory, which is
devices operating concurrently, time coincidence is no stored in ROM (read-only memory) or more widely in
problem. However, a microprocessor literally can EPROM (erasable programmable read-only memory).
handle only one task at a time. Multiplexors can The program stored in ROM or EPROM is non-
sample only one quantity at a time so voltages and volatile.
currents are not time-coincident. The awesome task of Another vital element in the architecture required
the programmer is to accommodate these peculiarities for microprocessor relaying applications is the

96 Chapter 6

Figure 6-1 Simplified diagram of a typical microprocessor relay.

NOVRAM (nonvolatile RAM) or EEPROM (electri- time processes, particularly in relaying applications
cally erasable programmable read-only memory). Data where so much is dependent on the time relationship
that are stored in this type of memory are not lost between quantities. There are two basic methods of
when power is removed from the relay. Settings and sampling data, both of which use a sample-and-hold
target data are usually stored here. circuit. An example of one method is shown in
Microprocessor-based algorithms typically require Figure 6-3. Using separate sample-and-hold circuits
time-coincident sampling of the input quantities. for each input, the microprocessor directs a ‘‘freeze’’ to
Considerable ingenuity is used to address this in real- occur at each sampling point. The S/H circuit holds

Figure 6-2 Typical sample and hold.

Microprocessor Relaying Fundamentals 97

Figure 6-3 Microprocessor relay with individual sample/hold.

this sampled value until the microprocessor can read in or voltage will be perceived by the relay considerably
each value through the multiplexor and A/D circuit. differently from its actual continuous waveform. High
The microprocessor then directs the S/H circuit to frequencies in the waveform cannot only fail to be
resume the sampling process until the next freeze identified due to inadequacies in the sampling process,
signal. but may indeed present themselves as a lower-frequency
An alternative sampling method that is less expen- component. Once this error intrudes into the process, it
sive is to use a single S/H circuit for all inputs. This is cannot be reconstituted and removed. Either the error
shown in Figure 6-4. A time correction factor is must not be allowed to occur in the first place by
applied to each sample after the first in a group. filtering out the offending frequencies or a process such
It is known precisely what difference in sampling as asynchronous sampling must be used. The mechan-
times exist and therefore all the samples in a time- ism of a high-frequency component in an input wave-
sequence-sampled (multiplexed) group can be con- form manifesting itself as a low-frequency signal is
verted to coincident samples by applying an angle called aliasing. It will now be described using a phasor
correction. Of course, other methods can be used. technique that may clarify the concept for the reader.


Because of the practical limitation of sampling rates in Figure 6-5 depicts the representation of a simple
a numerical relay, a varying input such as an ac current sinewave by a phasor as it rotates. The projection of

Figure 6-4 Basic hardware for microprocessor relay.

98 Chapter 6

It can be seen that the normal circle for the

fundamental is distorted into an ellipse by the presence
of the harmonic. Thus, when the total waveform is
constructed by the projection on the vertical, it can be
seen to be deficient in magnitude. This is the
phenomenon known as, aliasing. It is the appearance
of a high-frequency signal as a lower-frequency signal
that distorts the desired signal.

Figure 6-5 Generation of a sine wave by a phasor. 4 HOW TO OVERCOME ALIASING

4.1 Antialiasing Filters
the phasor on the vertical axis, at a given time,
This effect may be removed by filtering the high-
represents the magnitude of the sine wave at that time.
frequency components from the input. The element
Note that to do this the phasor must be represented by
that accomplishes this function is called an antialiasing
its peak value, not its rms value. Phasors are generally
filter. The ‘‘Nyquist criterion’’ states that in order to
manipulated, however, using rms values. They are
avoid the aliasing error, frequencies above one-half the
generally shown for a single frequency. Figure 6-6
sampling rate must be removed. Figure 6-8 shows a
extends this to show a fifth harmonic phasor super-
typical antialiasing filter.
imposed on the fundamental and the distorted sinusoid
that is generated by the combination. The fifth
harmonic phasor, of course, rotates through 4508 in 4.2 Nonsynchronous Sampling
the time required for the 60-Hz fundamental phasor to
rotate through 908. When the luxury of time exists in the relay response, an
Now the effect of high-frequency distortion of the alternative to the antialiasing filter can be used.
waveform on the sampling process can be examined. Nothing is lost in shifting the sampling points for the
Figure 6-7 uses an example of a seventh harmonic and fundamental frequency component of the quantity
an 8-per-cycle sampling rate. being measured. For example, in Figure 6-9, it is not

Figure 6-6 Phasor representation for fundamental and fifth harmonic.

Microprocessor Relaying Fundamentals 99

Figure 6-7 Aliasing effect of seventh harmonic.

critical that the first sample be taken at the zero the presence of the harmonic, for the first fundamental
crossing. It may be taken at any arbitrary point with cycle (see Fig. 6-7), it will for the second cycle
the following samples being equally spaced at the (following the jump) appear to be too high by the
sampling rate. If then, after collecting eight samples in same amount. Thus, a comparison of information in
this case, a jump is introduced to delay the beginning the adjacent cycles allows the effect of the seventh
of the collection of the next eight samples, by a time harmonic to be removed.
corresponding to 1808 of a particular high-frequency If the speed requirement of the device allows, then it
quantity (25.718 on a 60-Hz basis for the seventh is possible to eliminate the error associated with a
harmonic), an interesting effect takes place. If the particular harmonic without an antialiasing filter.
measured quantity appears to be too low, as a result of Note, however, while other harmonic frequency
components in the input signal are attenuated by this
asynchronous sampling procedure, the effect of only
one frequency is eliminated. This concept has been
used successfully in the MCO and MMCO relays.


With electromechanical relays, the designer has little

choice with regard to whether electrical quantities are

Figure 6-8 Antialiasing filter. (a) Diagram; (b) frequency

response. Figure 6-9 Waveform samples taken from a sinusoid.
100 Chapter 6

to be interpreted in terms of peak value, average value, tion coupled with a similar one using a cosine function
rms value, or fundamental frequency value. With the instead of a sine allows the complete magnitude and
power of the microprocessor, the designer can apply phase position of the desired frequency to be obtained.
any of these measurement techniques. With this integration, any desired frequency, such as
a 60-Hz fundamental, can be extracted from a
distorted periodic waveform, and all other frequencies
5.1 rms Calculation will be excluded.
The digital determination of root-mean-square values
of waveforms is quite similar to the conventional 5.3 Fourier-Notch Filter
analytical methods. Equations (6-1) and (6-2) illustrate
this. By squaring the magnitude of each sample (In) The comparable digital process involves the multi-
over a cycle and summing that with other squares, plication of individual samples by stored values from a
dividing the sum by the number of samples, and taking reference sine wave and summing the products over a
the square root, an rms value can be extracted from a full cycle
complex waveform.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi AC ¼ f K ðtÞCAK
Z 2p
1 K¼0
Analog rms ¼ I2 sin2 ot dt ð6-1Þ
2p 0 m X
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi AS ¼ f K ðtÞCBK
1 X8
Digital rms ¼ n¼1 n
I2 ð6-2Þ K¼0
Time overcurrent devices that are to be coordinated 2 K
with other apparatus which experience I2 R heating CAK ¼ cos 2p
effects (such as fuses, conductors, and transformers)  
2 K
have been developed with an rms response. For CBK ¼ sin 2p
applications in which harmonic effects are generated
by apparatus such as six-pulse rectifiers, it may be K ¼ number of sample
required that the harmonics be ignored. Relays have N ¼ samples per cycle
been designed using microprocessor techniques that where
are responsive only to the fundamental frequency
component of the input waveform. f(t) ¼ original function
T0 ¼ period of the waveform
n ¼ order of the harmonic
5.2 Digital Filters
Consider, for example, an application in which there
are eight samples per (fundamental 60-Hz) cycle. The
Any periodic waveform can be represented by a
corresponding samples of a sine wave may be chosen
fundamental and series of harmonic frequencies. Any
as 0, 0.707, 1.0, 0.707, 0,  0.707,  1.0, and  0.707.
particular frequency can be extracted by utilizing
These are fixed values, being sinðK2p=8Þ, where K are
Eqs. (6-3) through (6-5):
the samples 0 to 7 and 2p=8 corresponds to 458. These
Z T0 =2 values are then multiplied by 2/N to obtain the
an ¼ fðtÞ cos not dt ð6-3Þ constants that are used.
T0 T0 =2
Z T0 =2 If then the sampled values of the measured quantity
2 over a full cycle are multiplied by these constants in the
bn ¼ fðtÞ sin not dt ð6-4Þ
T0 T0 =2 proper order and summed, the process of Eqs. (6-3)
f n ðtÞ ¼ an cos on t þ bn sin on t ð6-5Þ and (6-4) is duplicated. This provides information that
excludes all frequencies except the fundamental. By
The sum of the product of the function and the sine of this process, any frequency component can be isolated
the frequency that is to be extracted, taken over the and utilized to perform a desired function. From this
period of the fundamental, produces a total that process results a function As ¼ K sin ot.
contains only the desired frequency. This is the Similarly, by using a set of cosine function
fundamental premise of Fourier analysis. This calcula- constants, the sample multiplication and summation
Microprocessor Relaying Fundamentals 101

can generate a function Ac ¼ K cos ot. Since input, it also is severely attenuated by this process of
sin2 ot þ cos2 ot ¼ 1; A2s þ A2c ¼ K2 . Thus, the peak summing.
value of a particular frequency component can be Using four samples per cycle, this digital filter
found by taking the square root of A2s þ A2c . Also since removes all of the even harmonics. With 60-Hz
sin ot= cos ot ¼ tan ot; As =Ac ¼ ðK sin otÞ = ðK cos otÞ waveform samples taken at 908 intervals, the samples
¼ tan ot: of 120 Hz would occur at 1808 intervals. Every other
The angle of the function can be found by sample will be equal, so S1  S3 ¼ 0 and S2  S4 ¼ 0.
  The second harmonic is eliminated by the summation
1 As Ss ¼ S1  S2  S3 þ S4 . With the fourth harmonic, the
y ¼ tan
Ac samples are taken at 3608 intervals and being equal
This algorithm is called a Fourier-notch filter. also produce Ss ¼ 0. All even harmonics are elimi-

5.4 Another Digital Filter 5.5 dc Offset Compensation

Other forms of digital filtering are used for specific Direct current offset in the fault current occurs as a
applications. In the IMPRS series of relays, four result of two natural laws: (1) Current cannot change
samples per cycle are used. These samples may begin at instantaneously in an inductance and (2) current must
any point in the cycle, such as at angle f in Figure 6-9. lag the applied voltage by the natural power-factor
The values of the individual samples can be described angle of the system. dc offset produces no desirable
as effects in overcurrent or distance relays. To make these
devices responsive only to the ac component of fault
S1 ¼ sinðot þ fÞ current, it is necessary to remove the dc by some
S2 ¼ sinðot þ f þ 90 Þ expedient.
S3 ¼ sinðot þ f þ 180 Þ The maximum dc component of the fault current is
S4 ¼ sinðot þ f þ 270 Þ Im ð1  et=T Þ, where Im is the peak value of the
symmetrical ac fault current, t the time in cycles, and
S5 ¼ sinðot þ fÞ T the dc time constant of the circuit that limits the fault
Digital filtering can be accomplished with the follow- current. The dc removal algorithm can be exact if T is
ing procedure: known. Unfortunately, for a given system it is likely to
pffiffiffi vary considerably.
Ss ¼ S1  S2  S3 þ S4 ¼ 2 2 sinðot þ f  45 Þ Many algorithms have been used. One uses the
concept that a sample of the fundamental component
Similarly, of current has the same magnitude as, and the opposite
pffiffiffi sign to, a sample taken 1808 later. The dc components
SC ¼ S2  S3  S4 þ S5 ¼ 2 2 cosðot þ f  45 Þ
for each of these samples are the same (if we assume
These values were related to the simple sine wave of this component is truly dc). Thus, for a 480-Hz
Figure 6-9 with a peak value of 1.0 for a sine wave, sampling rate
Im sinðot þ fÞ. The value of Im , the peak value, can be IK þ IK4
obtained by Offset ¼
S2s þ S2C where IK is the value of a sample of current and IK4
Im ¼ pffiffiffi the value taken four samples previously. With eight
2 2
samples per cycle, these samples would be 1808 apart
This digital filter also has other useful qualities. If we and the effect of the sinusoidal component would be
consider a dc current, it is obvious that S1  S2  S3 þ nullified in the summation. The offset may then be
S4 ¼ 0 because each sample is the same magnitude. used as a correction factor for the samples taken in this
Similarly, a linearly decaying ramp wave shape will interval.
also produce a sum equal to 0. Since the dc component With a decaying dc as opposed to an unvarying
of fault current has an exponential decay which is value, some error is introduced in this process,
between the constant dc input and linearly decaying depending on the dc time constant.
102 Chapter 6

5.6 Symmetrical Component Filter 508 shift in the sampling interval at S1 and S4 . This
assures over several cycles that a reasonable distribu-
Another interesting digital filter, utilizing three samples tion of samples is obtained and an accurate measure-
per cycle, is embodied in the MPR relay. In most ment of negative sequence current is achieved. This
applications, time-coincident quantities are necessary, algorithm is useful for long-term effects such as motor
but as the following symmetrical component defini- heating, but is unsatisfactory for fault detection. The
tions suggest, quantities that are 120 or 2408 displaced accuracy of this method is dependent on the nature of
in time are useful the harmonics.
IA1 ¼ ðIA þ aIB þ a2 IC Þ
3 5.7 Leading-Phase Identification
IA2 ¼ ðIA þ a2 IB þ aIC Þ
3 An interesting task for a microprocessor is to take a
1 collection of nonsimultaneous samples of voltages and
IA0 ¼ ðIA þ IB þ IC Þ
3 currents, determine the related phasors, modify them
as the algorithm dictates, and compare the resulting
where time-coincident phasors to establish which leads the
a ¼ 1ff120 other

The normal analog process for extracting IA2, for A ¼ ax þ jay

example, from the three-phase currents is to rotate IC B ¼ bx þ jby
by 1208 and IB by 2408, and add both to IA. With ð6-6Þ
ay bx ax by
digital techniques, an alternative procedure can be sin g ¼ 
used. A sample of IA is added to a sample of IC that is jAjjBj jAjjBj
taken 1208 later, then the sum is added to IB taken 2408 Equation (6-6) shows the concept that has been
later, giving the instantaneous value of 3IA2 that developed for digital relays following many years of
existed at the time of the IA sample. Figure 6-10 experience with various analog devices with similar
illustrates this process. A similar procedure is used to functions. It states simply and remarkably that phasor
extract 3IA1 from the individual samples of IA, IB and A leads phasor B if the product ay bx is more positive
IC that are taken at 5.55-ms (1208) intervals. Note the than ax by .
Only the difference in products, ay bx  ax by , is
required to determine which phasor leads the other.
No divide function, no sine or tangent calculation, and
no table lookup are required, thereby providing very
efficient use of the microprocessor. If g, the angle by
which phasor A leads phasor B, is between 0 and 1808,
the sine of g is positive. With jAjjBj being always
positive, sin g is positive if ay bx is more positive than
ax by .

5.8 Fault Detectors

Fault detectors provide an additional level of security

in a relaying application. They have been traditionally
of the simple-phase overcurrent variety with the
frequent addition of ground overcurrent. When the
fault current/load current ratio is small, compromises
must be made in the settings. Modern technology
allows more refined ‘‘fault detection.’’
Figure 6-10 Sum of samples taken 1208 apart equivalent to Phase current change, DI, and phase-to-ground
IA þ a IC þ a2 IB taken at S1 . voltage change, DV, can be implemented simply by
Microprocessor Relaying Fundamentals 103

6.2 Analog Test

Periodically, a known value of voltage is substituted

for the normal inputs to the multiplexer. The output of
the A/D converter is then checked for agreement with
the known input. If there is disagreement, an alarm
output results. If there is agreement, the multiplexer,
sample-and-hold, and A/D converter are proven to be
in good working order.

Figure 6-11 D1 fault detector. 6.3 Check-Sum

Any memory segment that is unchanging such as the

using a comparison of samples at similar points in ROM can be checked through the process of adding up
adjacent cycles as shown in Figure 6-11. These the contents of the memory and periodically verifying
changes, along with a change in zero sequence current, that the sum is fixed. Any change in the contents of the
provide a clear indication of the need for a distance ROM following power-up constitutes a failure and will
or directional unit to make a decision, and therefore produce an alarm.
they are used to shift some relaying systems (MDAR,
REL-301) to ‘‘fault mode.’’
By using DI (phase or I0 current change), a relaying
system can discriminate between loss of voltage caused 6.4 RAM Test
by a fault and that caused by potential circuit
problems. DI is present for faults and not loss of Random access memory is completely checked during
potential. the initializing process when power is applied to the
relay. Word patterns are written and read. Any
inconsistencies are identified.

6.5 Nonvolatile Memory Test
The ability to monitor much of the hardware is an
inherent part of any relaying system equipped with a Some relays utilize nonvolatile memory for storing
microprocessor. Some contain a provision for identify- details that are pertinent to the operation of the relay,
ing specific types of failure, whereas others indicate but will be changed by the user from time to time. An
only the general condition of check-failure, requiring a example of this is the settings. By storing the settings in
more detailed examination to pinpoint the nature of three locations when they are first entered and
the failure. comparing these three periodically, assurance is
obtained that they are correct. Inconsistency produces
an alarm.
The ability to test themselves is one of the principal
advantages of microprocessor relays. It relieves the
6.1 Dead-Man Timer need to apply external quantities to them periodically
to verify their capability to perform their intended
As part of the housekeeping tasks that are generally function. At the same time, it should be recognized
performed, a dead-man timer (also called watchdog that no relay is able to completely test itself in all
timer) supervises the fact that the microprocessor is respects. Backup relays are still required even though
cycling. If the microprocessor fails to perform a given those failures that do occur in microprocessor relays
function within a predetermined band of time limits, have a high probability of being identified immedi-
an alarm output is produced. ately.
104 Chapter 6

7 CONCLUSIONS in new approaches to old techniques, as well as

encouraging new innovative methods of solving
The introduction of microprocessor technology into persistent protection problems.
protective relaying has afforded us the ability to The microprocessor has established its place in
achieve new functions and self-checking provisions protective relaying and will occupy a position of
not previously possible. At the same time, it has caused prominence in future designs.
a reevaluation of long-established practices, resulting
System Grounding and Protective Relaying
Revised by: W. A. ELMORE

1 INTRODUCTION power system or where current can reasonably flow in

the earth structure.
Ground fault protection is dependent on the power- This chapter will cover protective relaying schemes
system grounding, which can vary from solidly for noneffectively grounded systems. These systems fall
grounded (no intentional impedance from the system into one of three categories:
neutrals to ground) to ‘‘ungrounded’’ (system
grounded only through the capacitance of the system).
Ground relaying for effectively grounded systems is
discussed in Chapter 12. In these systems, the X0/X1
ratio is 3.0 or less, and the R0/X1 ratio is 1.0 or less at In addition, this chapter will discuss the special
all points and under all operating conditions. With problems of sensitive ground relaying on distribution
effective grounding, the line-to-ground fault current is circuits and ground fault protection for both
equal to or greater than 0.6 times the three-phase fault ungrounded and multigrounded three-phase, four-
current. wire systems.
Solid grounding is necessary to meet these standard
criteria, particularly with overhead lines where the
X0/X1 ratio averages between 1.6 to 3.5. In solidly
grounded systems, the neutrals of the wye-delta power
transformers are directly connected to earth through 2.1 Ground Faults on Ungrounded Systems
the station ground mat. Considerable design effort is
expended to keep the resistance in this connection to a The term ungrounded is strictly one of definition,
minimum: Typical values of ground mat resistance to indicating no physical connection of any kind between
earth are on the order of 0.1 O or less in areas of low the system and ground. Since, however, there is always
ground resistivity. Typical values are higher in high distributed capacitance between the three phases of the
ground resistivity areas, resulting in a large station system and ground, the system is grounded through
ground mat rise (voltage gradient) between the station this capacitance. On such systems, current flows
area and remote grounds during ground faults. between each conductor and ground under normal
Earth, remote ground, and true earth are difficult conditions. In the event of a single line-ground fault,
terms to define precisely, since the earth is a very the corresponding line-to-ground capacitance is
heterogeneous mass. The terms represent a mathema- shunted out.
tical fiction needed to identify the zero potential earth Using symmetrical components, Figure 7-1 shows
plane. In practice, they are considered to exist within the networks and fault representation. Here, X1C, X2C,
the earth at any point remote from the influence of the and X0C are the total distributed capacitances of each

106 Chapter 7

Figure 7-1 Sequence network interconnection for ‘‘a’’

phase-to-ground fault on an ungrounded system.

system. Although they are shown here as a lumped

quantity, they are actually distributed parameters. X1C
and X2C can be neglected because their effect is
Figure 7-2 Ground fault on ungrounded system.
insignificant compared to that of X0C. X0C predomi-
nates so that approximately
at a high-frequency current zero, the trapped charge on
Ia ¼ 3I0 ¼ ð7-1Þ the line produces a voltage of 3.0 per unit. If a second
X0C restrike occurs at the next voltage positive maximum,
the peak voltage will overshoot to þ5 per unit as it
Since the ground fault current returns through the
shunt capacitance, the unfaulted phase currents are not
0 (Fig. 7-2). The phase-b and -c voltagespareffiffiffi shown as
the prefault line-to-line voltages or 3VLG . This
relation holds true only for the steady-state condition
with zero fault resistance; transient voltages can be
considerably higher as shown in Figure 7-3.
When the circuit breaker opens and extinguishes the
arc at or near current zero, the voltage is near its peak
pffiffiffi This voltage, shown in Figure 7-3 as 1.0 per unit
( 2 times the rms value), remains on the line (or right-
hand) side of the breaker, while the generator voltage
goes to the maximum negative value one half-cycle
later. At that time, the voltage across the breaker
contacts is essentially 2.0 per unit, the crest value. A
voltage of this value can cause the arc to restrike across
the breaker contacts, sending the line voltage from þ1
per unit to 1 per unit.
The result is a high-frequency transient voltage,
whose first peak overshoots the 1.0 value by 2.0
(the difference between 1 and þ1), giving a peak
voltage value by 3.0. If the arc is again extinguished Figure 7-3 Overvoltage due to reignitions and restrikes.
System Grounding and Protective Relaying 107

goes from 3 to þ1 per unit. Theoretically, further Users have successfully applied resistors across the
cycles of extinguishing and restriking of the arc would break in the broken delta configuration (as in Fig. 7-4)
build higher and higher voltage values. In practice, having a value that will limit current in the delta loop
however, flash-overs usually occur before these high to the rated current of the voltage transformer
values are reached. secondaries. Much higher values of resistance have
The peak voltage values shown in Figure 7-3 are also been used successfully. Karlicek and Taylor in
maximum theoretical values based on arc extinction at their important paper, ‘‘Ferroresonance of Grounded
zero current, no damping, and arc restrike at the crest Potential Transformers on Ungrounded Power Sys-
value of the source voltage. In fact, circuit resistance tems’’ (AIEE Power Apparatus & Systems, August
will introduce damping of the transient, reducing the 1959, pp. 607–618), concluded that the appropriate
peak value of the first half-cycle overshoot. Further, value of the resistor (called 3R in Fig. 7-4) is 100
restrikes may occur before the voltage reaches crest La/N2, where La is the voltage transformer primary
value voltage, which will reduce the value of peak inductance in millihenries and N the transformer turns
overvoltages. Nevertheless, overvoltages can be very ratio.
high and represent the major disadvantage of Typical resistor values in use are as follows:
ungrounded systems. An initial fault can cause a
second ground fault to occur on a different phase
possibly on a different feeder, producing a phase-to- Voltage transformer ratio Resistor (O)
phase-to-ground fault with its associated high current 2400–120 250
and damage. 4200–120 125
7200–120 90
2.2 Ground Fault Detection on Ungrounded 14,400–120 60

Since the fault current for a single line-ground fault on Although the primary fault current may be low,
an ungrounded system is very small, overcurrent relays high secondary currents can flow. This should be
cannot be used for fault detection. Voltage relays will checked with the short time or continuous rating of the
detect the presence of the voltage unbalance produced voltage transformer and resistor.
by the fault, but will not selectively determine its Applying a grounded-wye-broken-delta transformer
location in the system. The unbalanced phase and zero with only a relay connected across the break and no
sequence voltages that occur during ground faults are shunt resistor is equivalent to very-high-impedance
essentially the same throughout the system. Since grounding. Any shunt resistor, even as high as 20Xc, is
selective isolation of the fault is not possible, relay better than none. It will damp any high transient
schemes are only useful for providing an alarm. voltage oscillations and probably hold the peak values
Figure 7-4a shows the preferred ground fault to less than twice the normal crest voltage to ground.
detection system. The voltage transformers must have The alternative ground fault detection scheme (Fig.
a primary voltage rating equal to the line-line voltage, 7-4b) is not recommended and should only be applied
since this is the voltage that will be impressed on the after careful study. The CVD electromechanical relay
two unfaulted phases during a line-ground fault. or the type 27/59 solid-state relay in this system has
Under normal conditions, the voltage across the relay separate contacts for operation on either over- or
is approximately 0. When a single line-ground fault undervoltage. With the vt connected to phase c, line-
occurs, the voltage becomes 3V0, or approximately to-ground faults on phases a and b produce an
200 V with 69-V secondary windings. The electro- overvoltage on the relay; faults on phase c produce
mechanical type CV-8 relay or solid-state type 59G an undervoltage. For the scheme to work, the
relay shown, with their 200-V continuous rating, will capacitance to ground of the lines must be fairly
detect fairly high-resistance faults. closely balanced and high enough to keep the neutral
Wye-connected transformers with grounded neutral of the system at close to ground potential.
on an ungrounded system may be subject to ferror- This scheme can also produce ferroresonance or
esonance during switching or arcing ground faults. To neutral inversion. When XC/XL is 3.0, VT theoretically
avoid this, care must be exercised in planning the would be infinite. Even without faults on the system,
relationship between the magnetizing impedance of the the high magnetizing impedance of the voltage
transformer, its knee point voltage, and its load. transformer can approach resonance with the line
108 Chapter 7

Figure 7-4 Ground fault detection on ungrounded systems.

capacitance to neutral, causing a high overvoltage ground faults in the generator. As a result, the faults
across the secondary. Neutral inversion can occur persisted, causing undue damage.
during a line-ground fault on a phase other than c. The initial solution was to connect the generator
Such a fault produces unbalanced impedances to neutral to ground through the primary of a voltage
ground; the resultant current flows can drive the transformer and put an overvoltage relay across the
system ground point outside the delta. A loading secondary. In theory, a single line-ground fault would
resistor across the relay or, less desirably, in series with simply cause the generator neutral voltage to shift with
the transformer primary may prevent these problems. respect to ground, activating the relay and tripping the
machine or sounding an alarm. In practice, however,
this system actually increased the machine failure rate.
The cause was arcing grounds—a phenomenon similar
to the restrikes that can occur when switching a
capacitive reactance.
There are three different types of reactance grounding:
The arcing ground phenomenon can be explained
High-reactance grounding using Figure 7-5. The equivalent single-line diagram
Resonant grounding shown in Figure 7-5a is for a generator grounded
Low-reactance grounding through a high reactance Xn, with a line-to-ground
fault near one terminal. Xc is the distributed capacitive
reactance of the windings to ground, connected half-
3.1 High-Reactance Grounding way between the generator reactance Xg. If the arc is
extinguished when the small fault current passes
Until the early 1940s, some utilities operated their unit- through 0, the voltage across the arc path must go
connected generators with the neutral ungrounded. from nearly 0 to the normal crest value. In doing so, it
Their purpose was to keep the internal line-to-ground must oscillate around the steady-state normal value.
fault current in the generator very low and prevent the As shown in Figure 7-5b, the resultant voltage
iron from being damaged by arcing. Unfortunately, the transient will reach a peak value of twice the normal
result was a high insulation failure rate in machine crest line-to-neutral voltage, one-half cycle of the high-
windings. frequency transient after the arc is extinguished. If the
These failures were caused by high-voltage transi- arc restrikes at this point, the fault voltage is driven
ents, similar to those discussed earlier. This problem back to 0. When the arc is initially extinguished, the
was compounded by an inability to detect single line- reactor voltage has to go from the positive maximum
System Grounding and Protective Relaying 109
pffiffiffi pffiffiffi
pffiffiffi theoretical transient peaks are 3; 3 3;
5 3, and so on. For these reasons, high-reactance
grounding was discontinued many years ago.

3.2 Resonant Grounding (Ground Fault


In certain sections of the United States, resonant

grounding has been applied successfully in unit-
connected generator grounding applications. It is not
applied in transmission line applications in the United
States, but other countries use it. In this scheme, the
total system capacitance to ground is compensated for
or cancelled by an inductance in the grounded neutral
of the power transformers. The grounding reactor,
equipped with taps that permit it to be tuned to system
capacitance, was first called a Petersen coil. It is now
more commonly designated a ground fault neutralizer.
Theoretically, if the reactor perfectly matches the
system capacitance, a line-to-ground fault will produce
zero current, the transient fault arc will be extin-
Figure 7-5 Overvoltages on reactance grounded system due guished, and the arc path deionized, without the need
to arcing fault. for deenergizing the circuit.
In this system, approximately 75% of line-ground
faults are self-extinguishing. The remaining faults must
to 0. As a result, it has a transient oscillating period be cleared by a line breaker.
from the positive maximum to negative maximum. In theory, resonant grounding should reduce line
The first half-cycle of this oscillation is shown in outages considerably. This system does, however, have
Figure 7-5c. If the arc restrikes at the instant when the a number of disadvantages:
fault voltage is twice the normal crest value, as was Transformers connected to the system must have
assumed in Figure 7-5b, the reactor voltage has to go full line-line insulation even when wye-con-
from the negative minimum to positive maximum. The nected.
result is another transient oscillation, with a peak value The entire system must be fully insulated for line-
of three times the normal maximum line-ground line voltage.
voltage. The ground-fault neutralizer must be retuned to
Note that in high-reactance grounding, the reactor accommodate any changes in system configura-
voltage is applied between the generator neutral and tion: additions, extensions, line removals, or
ground. Since the BIL of the reactor is higher than that switching.
of the generator windings, insulation failures are more System effectiveness will be reduced considerably if
likely to occur in the generator windings. a substantial number of lines are of wood pole
The switching surges that result from clearing line- construction. The high insulation to ground will
to-ground faults for ungrounded systems also occur in result in a larger portion of line-line faults
high-reactance grounded systems. In the latter case, the (conductor swing caused by wind).
resulting transient overvoltages will be even higher. A high incidence of faults will occur essentially
The source voltage for an ungrounded system is the simultaneously in different parts of the system.
normal line-to-neutral voltage that, theoretically,
produces successive line-side voltage peaks of 1.0,
3.0, 5.0, . . . of normal crest voltage to neutral. For the 3.3 Low-Reactance Grounding
high-reactance grounded system shown in Figure 7-5,
with the reactor between the neutral and ground, the Low-reactance grounding used to be applied to
source voltage is the normal line-line voltage. The systems fed at generator voltage. The generator neutral
110 Chapter 7

was grounded through a reactor. The reactor was sized circuit value, if we assume that X2 ¼ X1. Thus, if a
to keep the magnitude of a single line-to-ground fault ground relay is used in the common neutral connection
on the machine terminals equal to a three-phase fault. of the line current transformers, the current level is a
[A reactor value of (2X1  X0  X2)/3 was used.] maximum of half that of the phase current for a three-
In general, low-reactance grounding was applied to phase fault.
large industrial plant systems with radial distribution
feeders, and ground fault protection consisted simply
of overcurrent relays. Gradually, this type of generator
grounding has been replaced by low-resistance ground-
Resistance grounding is applied in systems with three-
Another type of low-reactance grounding provides
wire distribution at the generator voltage and for unit-
ground fault current relaying for systems supplied
connected generators. The two general types of
from a delta source. The reactance grounding scheme
resistance grounding are low- and high-resistance
Supply sufficient ground fault current to operate
relays for a fault when the line value (X0 þ 2X1) is
4.1 Low-Resistance Grounding
the highest.
Limit the transient overvoltages attributable to
Whenever low impedance grounding is desired, resist-
ground faults to a value of 2.5 times normal
ance grounding is generally preferred to the low-
line-to-neutral crest value, with two restrikes.
reactance systems described above. Specifically, low-
In this scheme, either a grounded wye-delta or zig-zag resistance grounding is used for systems fed directly at
transformer can be used, although the zig-zag (Fig. the generator voltage (Fig. 7-7a) or through a delta-
7-6) is more common because of its economy. The wye transformer (Fig. 7-7b). When a line-to-ground
windings shown parallel are on the same core leg. With fault occurs in the system, the current flowing in the
this connection, the positive sequence impedance of the ground resistor results in a sudden change in generator
bank is very high and equal to the magnetizing
impedance. When zero sequence current passes
through the bank as shown, the impedance is equal
to the leakage reactance.
The rating of the transformer is chosen so that the
maximum X0/X1 value is 4. When X0/X1 equals 4, the
line-ground fault current is half the three-phase, short-

Figure 7-7 Low-resistance grounding of systems fed

Figure 7-6 Reactance grounding. through Delta-Wye transformer.
System Grounding and Protective Relaying 111

load, causing severe generator angular swings and 4.2 High-Resistance Grounding
high-peak shaft torques. To keep the ground resistor
loss low, the resistor is generally sized to limit the High-resistance grounding is applied to a generator-
single line-to-ground bus fault to around 100 to 400 A. transformer unit system by connecting a resistor across
Ground relaying using residual overcurrent relays may the secondary of a distribution transformer in the
be applied, but zero-sequence-type current transfor- grounded generator neutral (Fig. 7-8). The resistor
mers will provide greater sensitivity. value is selected so that its KW loss for a solid line-to-
For distribution systems, transient overvoltages ground fault at the machine terminal is equal to or
may be limited to less than 2.5 times the normal crest greater than the charging kVA of the low-voltage
value to ground, if system. To do this, 3R is chosen to be equal to or less
than Xco. Xco is the combined zero sequence
R0 X0 capacitive of the generator windings, cable connections
 2:0 and  20 ð7-2Þ
X0 X1 to the transformer, low-voltage transformer winding,
and station service transformer plus any surge protec-
For a neutral resistor R, tive capacitors that are applied at the generator
terminals and connected phase to ground. This resistor
R0 ¼ 3R value will limit generator iron burning from ground
faults, damp out oscillations, and limit the peak
Assuming that X0 ¼ 20X1 and R0 ¼ 2X0 , then we have transient voltage to 2.5 times normal line-to-neutral
voltage, or less. The rating required for the resistor is
Z0 ¼ ð40 þ j 20ÞX1
For a line-to-ground fault, Resistor KWR ¼ ð7-6Þ
3 where V is generator-rated phase-to-neutral voltage (in
Ig ¼ 3I0 ¼
X1 þ X2 þ Z0 volts).
3:0 The transformer kVA requirement is the same as
¼ this resistor KW value, of course, but the kVA rating is
ðj 1 þ j 1 þ 40 þ j 20ÞX1
ð7-3Þ higher because of the choice of a higher-voltage rating
3:0 3:0
¼ ¼ ff  28:8  for the transformer to avoid possible ferroresonance.
ð40 þ j 22ÞX1 45:65 X1
¼ ff  28:0  per unit

The three-phase fault current would be

I3f ¼ per unit ð7-4Þ

Thus, if we use Eq. (7-4), the line-to-ground fault

current magnitude is

Ig ¼ 0:066 I3f ð7-5Þ

The resistor can be in the neutral of the transformer

(Fig. 7-7), or a resistor can be inserted in the neutral of
the grounded zig-zag transformer (Fig. 7-6). In either
case, the reactance component of the resistor must be
considered. Cast-iron grid-type grounding resistors
have a power factor of approximately 0.98, stainless
steel types one of approximately 0.92. The reactance,
while small in itself, is tripled in the zero sequence Figure 7-8 High-resistance grounding of the unit connected
circuit. generator-transformer.
112 Chapter 7

The magnitude of primary fault currents in these

applications is around 8 to 10 A.
Sensitive protection is provided by an overvoltage
relay across the resistor. This application is detailed
under generator protection (Chap. 8).


Figure 7-9 Ground protection with conventional current
transformers and protective relays.
Ground relaying on distribution circuits can be
difficult. The range of fault currents can vary from
negligible, for a conductor lying on or near the ground
with minimum electrical contact, to substantial, for a 5.1 Ground Overcurrent Relay with
conductor making good contact with ground. Unfor- Conventional Current Transformers
tunately, there is no practical way of distinguishing an
intolerable situation from a tolerable one at a breaker In the scheme shown in Figure 7-9, the relays are
or disconnection location. usually set on the 0.5-A tap. Because of the large
Some years ago, a utility conducted tests on a 10-ft burdens of electromechanical ground relays on the
length of no. 4 bare copper wire energized at 12 kV and minimum tap, the relay pickup current multiplied by
laid on a variety of surfaces such as dry grass, green the current transformer ratio will not be the primary
vegetation, dry base soil, and asphalt. Of 128 tests, 7% ampere pickup when using lower-quality current
showed currents of less than 7 A, 7% over 1000 A, and transformers (see Chap. 5). To hold the exciting
55% had currents in the range of 150 to 600 A. current to a reasonable minimum, it may be necessary
Ground fault protection is dictated by the amount to use a higher tap setting than would otherwise be
of ground fault currents available from the system to desired, a solid-state or numerical relay, or a better
operate relays and the ratio of this current to normal current transformer.
system residual unbalance. Load management may The unequal performance of current transformers
help to reduce normal unbalance in some cases. The during heavy phase faults or initial asymmetrical
minimum ground fault current must balance service motor starting currents may produce false residual
continuity with equipment protection. That is, it must currents with the scheme shown in Figure 7-9a. When
be low enough to minimize equipment damage, but these currents cause relay operation, an instantaneous
high enough to be recognizable and allow the faulted relay with a higher pickup should be substituted, or the
area to be selectively isolated without nuisance time overcurrent relay should have a larger time dial
tripping. and/or pickup setting. Increasing the burden on the
The design of the system grounding should be current transformers in these cases causes them to
compatible with the sensitivity of the relaying that is to saturate more uniformly, reducing the false residual
be used. Three commonly used ground relay schemes, current.
in order of increasing sensitivity, are Higher burdens, however, may also decrease the
relay sensitivity on light ground faults, depending on
Ground relay in the common neutral connection of the quality of the current transformers. False residual
the line current transformers and/or grounded currents do not occur in the scheme shown in Figure
source (Fig. 7-9). 7-9b or 7-12 and do not cause relay operations in
Ground relay in the common neutral connection of Figure 7-10a or b.
the line current transformers, with a product- With the application of a ground relay set on the
type relay to avoid operation on false residual 0.5-A tap, the fault current in the relay should not be
currents (Fig. 7-10). The CWP scheme (Fig. less than twice pickup, or 1.0 A secondary.
7-10b) provides increased sensitivity, whereas the To be able to detect a ground fault that produces a
CWC scheme will not (see also Fig. 7-11). current having a value of 10% of the current produced
Ground relay with a zero sequence (ring) type of by a bolted phase-to-ground fault is a reasonable
current transformer (Fig. 7-12). criterion. The maximum load on any circuit off a bus
System Grounding and Protective Relaying 113

Figure 7-10 Sensitivity ground protection utilizing product-type relays.

dictates the critical ct ratio. With a primary current of relaying scheme to be used, such as a zero sequence
0.10 IG, a ratio of K:5 and minimum secondary current (ring) type of current transformer with its low ratio
of 1 A produce the following requirement: (typically 50:5) and an instantaneous overcurrent
relay, such as an IT or 50D.
IG  2 K ð7-7Þ
That is, the current permitted by the grounding device
for a bolted phase-to-ground fault should equal or 5.2 Ground Product Relay with Conventional
exceed twice the primary rating of the largest ct used Current Transformers
on a circuit off of the bus. If the use of this criterion
produces an excessively high ground fault current, a Better security is possible using the schemes of
lower value of current can be chosen by using a higher Figure 7-10, and increased sensitivity may be provided
resistance value. This will then require a more sensitive

Figure 7-11 Phasors for Figure 7-10b for a phase ‘‘a’’-to- Figure 7-12 Ground protection utilizing the zero sequence
ground fault on a high-resistance grounded system. (ring) type current transformers.
114 Chapter 7

by the scheme of Figure 7-10b. These schemes will not degree of saturation. This relay and ct combination is
operate on false residual currents; the relays require intended to provide sensitive detection of ground faults
current in both windings to operate. No system ground and is not expected to perform adequately in the
or zero sequence current or voltage will exist for phase presence of fault currents beyond the moderate (up to
faults or motor starting currents. 1800 A rms symmetrical primary) range. Where larger
Two product-type devices are used in the scheme of fault currents are expected, the system of Figure 7-9a
Figure 7-10b: the CWP and CWP-1. The CWP is and a larger ct ratio should be used.
applicable for reactance grounded systems and the When the maximum fault current exceeds the
CWP-1 specifically tailored to high-resistance maximum values shown, the output waveform is
grounded systems with its 45 8 lead characteristic. nonsinusoidal. Relay timing will tend to become
The greatest ground fault sensitivity is provided by variable and longer than indicated in the published
the CWP-1 (32N) relay. The relay pickup is adjustable literature.
between 5 and 40 mA with 100 V across the potential The above schemes are for feeder circuit protection.
coil at maximum torque. For the ground protection of equipment, a ground
The phasors for a high-resistance grounded system differential scheme can be used with a differential-type
where the CWP-1 relay is applicable are shown in relay or product-type (CWC) relay as shown in
Figure 7-11. Figure 7-13. This is also applicable to short-run feeders
with three conventional ct’s or a zero sequence type ct
at each end of the protected zone. The CWC relay is
5.3 Ground Overcurrent Relay with Zero recommended as it provides high sensitivity and is
Sequence Current Transformers relatively independent of the current transformer
The scheme shown in Figure 7-12 provides maximum
sensitivity. There are no false residual currents. The
zero sequence type of current transformer has the
conductors passed through the center hole, and the ct
ratio is not dictated by the load current. Secondary
current is the transformed system zero sequence 6.1 Unigrounded Four-Wire Systems
current 3I0.
The standard ratio for the zero sequence type of Unigrounded, four-wire systems have insulated neu-
transformer (type BYZ) is 50/5; 100/5 ratios were trals; the only ground connection is at the substation.
originally used. Various nondirectional relays can be Loads generally are connected phase to neutral, and
applied, as outlined in Table 7-1. the net load unbalance returns through the neutral as a
The IT relay has a large burden when the 0.15-A tap residual current. For faults from phase to ground, the
is used (19.6 O). Saturation of the 50/5 ct will occur at current returns through the earth to the substation
roughly 5 A primary. Field tests indicate that the neutral.
secondary pulse width at 1800 A primary is only There are three different relay schemes for ground
approximately 30 8 following each zero crossing, but fault protection for unigrounded systems, as shown in
this relay operates satisfactorily with this extreme Figure 7-14. Figure 7-14a illustrates the conventional

Table 7-1 Relay Settings and Sensitivities Using the 50/5 BYZ Zero Sequence Current Transformers

Minimum sensitivity in primary 3I0 Maximum primary 3I0 amperes for

amperes accurate timing and coordination

Relay type Relay setting (43=4 ID) (73=4 ID) (43=4 ID) (73=4 ID)

IT 0.15 5.0 5.0 — —

CO-8 or 9 0.5 9.0 10.0 25 112
CO-8 or 9 2.5 24.0 24.0 540 1215
CO-11 0.5 6.0 7.0 70 150
CO-11 2.5 24.0 24.0 700 900
System Grounding and Protective Relaying 115

Figure 7-13 Ground differential for wye winding using

CWC (product-type relay).

scheme used on three-phase, three-wire systems. For a

four-wire system, the load unbalance current would
flow through the ground relay, requiring a setting to
avoid operation on the maximum load unbalance. This
scheme is generally not recommended for the uni-
grounded system. The four-current transformer
scheme shown in Figure 7-14b provides much higher
sensitivity, since it does not measure the load
Figure 7-14 Methods of ground protection on unigrounded
unbalance residual current. Even greater sensitivity is
provided by the zero sequence type of current
transformer depicted in Figure 7-14c. Comparative
sensitivities for various relays in this scheme are listed
ground return may be a high impedance path, causing
in Table 7-1.
low voltage at the load points, the more sensitive
If a line-to-neutral fault occurs on the system, only
window-type current transformer scheme is recom-
the conventional scheme (Fig. 7-14a) will respond. The
connection of the current transformers in the other two
schemes results in cancellation of the fault current,
unless it involves ground. The phase relays will provide 6.2 Multigrounded Four-Wire Systems
protection, however, since phase-to-neutral fault cur-
rent in one phase will be of the same order of Many three-phase, four-wire distribution systems are
magnitude as a three-phase fault. solidly grounded at the substation, with the neutral
A fault between neutral and ground is possible even wire also grounded at each distribution transformer
though the neutral is nearly at ground potential, location. Such systems are difficult to protect against
probably as the result of a broken neutral conductor. ground faults. The scheme of Figure 7-9 is used most
The schemes of Figures 7-14b and c will measure any often with the sacrifice of sensitivity dictated by
current returning through the earth. Because the maximum load unbalance.
Generator Protection
Revised by: C. L. DOWNS

1 INTRODUCTION were applied. The amount of protection that should be

applied will, of course, vary according to the size and
The frequency of failure in rotating machines is low importance of the machine.
with modern design practices and improved materials,
yet failures will occur and delayed tripping or
insensitivity of protection may result in severe damage 2 CHOICE OF TECHNOLOGY
and long outages for repairs. For these reasons,
abnormal conditions must be recognized promptly In the choice of relays to be applied for the various
and quickly isolated to avoid extending the damage or functions described here, it will be recognized that they
compounding the problem. are available as discrete functions in their individual
Abnormal conditions that may occur with rotating housing or as a complete complement containing all
equipment include the following: the pertinent protection plus data acquisition.
Electromechanical, solid-state, and microprocessor-
Faults in the windings
based devices are used depending on personal choice
and whether or not a new installation, an upgrade, or a
Overheating of windings or bearings
functional addition is involved.
Although electromechanical and single-function
Loss of excitation
solid-state relays have proven their reliability, flex-
ibility, and effectiveness, the trend is toward micro-
Inadvertent energization
processor-based integrated packages. Many of these
Single-phase or unbalanced current operation
provide event recording, oscillography, self-monitor-
Out of step
ing, communications, adaptive characteristics, and
Subsynchronous oscillations
other features that only a microprocessor-oriented
Some of these conditions do not require that the unit system can provide.
be tripped automatically, since in a properly attended
station, they can be corrected while the machine
remains in service. These conditions are signaled by 3 PHASE FAULT DETECTION
alarms. However, most require prompt removal of the
machine from service. Internal faults in equipment generally start as a ground
For any particular hazard, the initial, operating, in one of the stator windings and may occasionally
and maintenance costs of protective schemes and the develop into a fault involving more than one phase.
degree of protection they afford must be carefully Differential protection is the most effective scheme
weighed against the risk encountered if no protection against multiple-phase faults. In differential protec-

118 Chapter 8

tion, the currents in each phase, on each side of the an external fault. On a symmetrical basis (no dc offset
machine, are compared in a differential circuit. Any in the primary current), this would not exceed 10 A if
‘‘difference’’ current is used to operate a relay. a C class ct were used within rated burden and the
Figure 8-1 shows the relay circuits for one phase ratio were chosen such that the secondary current did
only. For normal operation or a fault outside the two not exceed 100 A for the maximum ‘‘through’’ phase
sets of current transformers, Ip entering the machine fault. To avoid operation on an asymmetrical fault,
equals Ip leaving the machine in all phases, neglecting the trip time would have to exceed three dc time
the small internal leakage current. The secondary constants.
current of each of the ct’s is the perfectly transformed
primary current minus the magnetizing current Ie.
The relay current Ie1  Ie2 is the difference in the 3.1 Percentage Differential Relays (Device 87)
exciting or magnetizing currents. With the same type of
current transformers, this current will be small at The percentage differential relay (Fig. 8-2) solves the
normal load. If a fault occurs between the two sets of problems of poor sensitivity and slow operation. The
current transformers, one or more of the left-hand inputs from the two sets of current transformers are
currents will suddenly increase, whereas currents on used to generate a restraint quantity. This is then
the right side will either decrease or increase and flow compared to the difference of these two currents.
in the opposite direction. Either way, the total fault Operation (or restraint) is produced as a result of
current will now flow through the relay, causing it to the comparison of the difference to the restraint.
operate. This desensitizes the relay for high external fault
If perfect current transformers were available, an currents.
overcurrent relay in the ‘‘difference’’ circuit could be The current required for relay operation increases
set to respond very sensitively and quickly. In with the magnitude of the through fault current. The
practice, however, no two current transformers will percentage of increase may be constant, as in the CA
give exactly the same secondary current for the same (87) generator percentage differential relay. Alterna-
primary current. Discrepancies can be traced to tively, the percentage of increase may vary with the
manufacturing variations and differences in secondary external fault current, as in the high-speed SA-1 (87)
loading caused by unequal length of relay leads and generator relay. The effect of the restraint on internal
unequal burdens of meters or instruments connected faults is negligible, because the operating quantity is
in one or both secondaries. The differential current weighted and responds to the total secondary fault
produced flows through the relay. Although normally current.
small, the differential current can become appreciable Generator differential relays are available with
when short-circuit current flows to an external fault. various percentage differential characteristics. They
An overcurrent relay would have to be set above the are typically 10%, 25%, and variable-percentage
maximum error current that could be expected during differential types. The percentage indicates the differ-

Figure 8-2 Schematic connections of the percentage differ-

Figure 8-1 The basic differential connection. ential type relay. (Only one-phase connections are shown.)
Generator Protection 119

ence current as a percentage of the smallest restraint

current required to operate the relay. The pickup (the
value of current into one restraint winding and out the
operating winding) is the current required to barely
make the relay operate. Its value tends to be smaller
for the lower-percentage differential relays and is as
low as 0.10 A for some. Operating time, in general, is
smaller for solid-state relays, being 25 msec for the
solid-state SA-1 compared to 80 to 165 msec for the
electromechanical CA relay.
Multifunction microprocessor relays do not have a
physical operating winding, the difference of the
restraint currents being computed mathematically by Figure 8-3 Percentage differential relay schematic for a
the protection algorithm. delta-connected machine. (Only one-phase connections are
In all differential schemes, it is good practice to use shown.)
current transformers with the same characteristics
whenever possible and avoid connecting any other
equipment in these circuits. 3.4 Split-Phase

Generators with split-phase windings can be protected

by two sets of differential relays: one connected as in
3.2 High Impedance Differential Relays Figure 8-2 and the other as in Figure 8-4. This
(Device 87) arrangement protects against all types of internal
phase faults, including short-circuited turns or open-
High impedance differential relaying is based on the circuited windings. This scheme may be extended to
conservative premise that the ct’s on one side of the accommodate other winding arrangements involving
generator perform perfectly for an external fault and more than two equal windings per phase. Unless the
the other set of ct’s saturate completely. It takes ratios of the current transformers produce an exact
advantage of the fact that the voltage appearing across match, the scheme of Figure 8-4 must be equipped with
the relay is limited for an external fault to the voltage auxiliary transformers to provide a balance during
drop produced by the maximum secondary current normal operation.
flowing through the leads from the relay to the
saturated ct and through its internal resistance. For
an internal fault, the voltage will approach the open-
circuited ct voltage (usually limited by a varistor
internal to the relay). In general, this scheme is not as
sensitive as the percentage differential scheme but is
more secure.

3.3 Machine Connections

Most generators have wye-connected windings. As

shown in Figure 8-2, three relays connected to wye-
connected current transformers provide phase and, in
some cases, (depending on the type of neutral and
system grounding) ground fault protection. Figure 8-3
illustrates a similar protective scheme for delta
generators. In this scheme, the delta windings must
be brought out so current transformers can be installed Figure 8-4 Schematic connections for one-phase only for
inside the delta. the protection of a machine with split phase windings.
120 Chapter 8

4 STATOR GROUND FAULT PROTECTION age of the winding, this relay is often referred to as a
95% relay.
The method of grounding affects the degree of
protection afforded by differential relays. The higher 4.1.2 Neutral Third Harmonic Undervoltage
the grounding impedance, the less the fault current
magnitude and the more difficult it is to detect high Other schemes take advantage of the presence of the
impedance faults. With high impedance grounding, the third harmonic voltage between neutral and ground
differential relays will not respond to single-phase-to- and respond to undervoltage for a neutral-to-ground
ground faults. A separate relay in the grounded neutral fault.
will provide sensitive protection, since it can be set
without regard to load current. 4.1.3 100% Winding Protection
The ground relay may also operate for ground Other ground relaying schemes provide complete
faults beyond the generator. For this reason, a time protection of the generator stator by injecting a signal
delay may be necessary to coordinate with any into the stator and monitoring it for change. This
overlapped relays. A typical case is a generator concept allows 100% coverage even though the
connected directly to a bus with other circuits. A fault machine is at standstill, whereas the 95% and neutral
on one of these circuits should not trip the machine; third harmonic schemes depend on the machine
the relays in the faulted circuit will clear such faults. A operating at rated speed and voltage.
wye-delta transformer bank will block the flow of
ground current, preventing faults on the opposite side
of the banks from operating ground relays. In the unit- 4.2 95% Ground Relays
connected scheme, the transformer bank limits the
ground relay operation to faults in the generator, the The CV-8 or solid-state 59G low-pickup overvoltage
leads up to the transformer bank, and the delta relay can be used for unit-generator applications as
winding. shown in Figure 8-5. Provided that a full-rated primary
winding is used, the pffiffimaximum
ffi voltage for a solid
ground fault is 120= 3 (69.3 V with apffiffi120-V
ffi distribu-
4.1 Unit-Connected Schemes
tion transformer secondary), or 240= 3 (138.6 V with
a 240-V secondary).
The unit-connected system is the most common
The scheme has good sensitivity for internal ground
arrangement for all but small generators. For unit
faults while being very insensitive to third harmonic
systems high-resistance grounding is used, and the
voltages. Various provisions are used to make the relay
machine is generally grounded through a distribution
insensitive to the third harmonic. The third harmonic
transformer and resistor combination, as shown in
pickup of the CV-8 relay, for example, is approxi-
Figure 8-5. Since the secondary is rated at 120 or
mately 8 times the pickup at rated frequency.
240 V, the physical size of the resistor can be somewhat
The voltage appearing from the neutral of the
smaller than if it were connected to the primary.
generator to ground is dependent on the location of the
ground fault. The more sensitive the ground relay, the
4.1.1 ‘‘95% Scheme’’
greater the percentage of the winding protected.
The unit system responds to the voltage shift of the Obviously, a neutral-to-ground fault goes undetected
generator neutral with respect to ground that occurs by this relay and other devices must be considered. In
for a ground fault in the machine, bus, or low-voltage some relays, the sensitivity is proportional to its rating,
winding of the transformer. The relay used must be and the best protection is obtained by using the relay
insensitive to the substantial normal third harmonic having the lowest voltage rating. For a ground fault at
voltage that may be present between neutral and the line terminal of the machine, full line-to-neutral
ground, and yet sensitive to the fundamental frequency voltage will exist from neutral to ground. The voltage
voltage that accompanies a fault. on the relay is, of course, dependent on the ratio
Since the magnitude of the neutral shift is dependent chosen for the distribution transformer. If this voltage
on the location in the winding of the ground fault exceeds the rating of the relay and is not removed by
(neutral-to-ground fault produces no neutral shift) and tripping the field circuit within the short-time cap-
the usual choice of relay sensitivity and distribution ability of the relay, the SV scheme shown in Figure 8-5
transformer voltage ratio provides roughly 95% cover- may be used to protect the relay for those fault cases
Generator Protection 121

Figure 8-5 Schematic connections for ground fault protection of a unit type machine resistance grounded through a
distribution transformer.

producing this high voltage. The SV relay is set to open Another scheme that is often used in industrial
its contacts at a value somewhat lower than the applications uses a product-type relay for ground
continuous rating of the protected relay, inserting R to detection. This arrangement is described in Chapter 7,
limit the voltage. In general, the 95% type relay is Figure 7-13. Good sensitivity is achievable because of
allowed to trip immediately to remove voltage prior to the ability to use a low-ratio ct in the neutral
the occurrence of damage to the relay, making the SV grounding resistor path that is not related to the
unnecessary. Other solid-state relays such as the type machine full-load current.
59G have a 208-V continuous rating and 60-Hz
sensitivity as low as 1 V.
Time-delay settings of 25 msec to 4 sec are used for
this function. These longer delays allow for coordina- 4.3 Neutral-to-Ground Fault Detection (Device
tion with voltage transformer fuses, if required. 87N3)
Operation of the ground relay can be avoided for
faults on the main voltage transformer secondary by Figure 8-6 describes a scheme for detecting a neutral-
grounding one phase of the secondary rather than the to-ground fault on the generator. This fault is, in itself,
neutral. Then a ground fault on the voltage transfor- not hazardous. A second ground fault at the machine
mer secondary will not produce a machine neutral terminal, however, causes a line-to-ground fault that is
voltage shift and the ground relay will not operate. not limited by any neutral impedance. This fault
This is the recommended grounding practice for current magnitude will quite likely exceed the current
generator voltage transformers. magnitude for which the machine is designed. Machine
122 Chapter 8

Figure 8-6 180-Hz voltage comparator.

destruction may result. Early detection, then, is 4.4 100% Winding Protection
The scheme of Figure 8-6 compares the third Another important ground fault detection scheme
harmonic voltage present between the machine neutral (GIX-104) utilizes an injected current that is monitored
and ground with that at the line terminals. The relative for magnitude. In the configuration shown in Figure
values of these voltages are established by the 8-7, the small injection voltage is applied across the
distributed capacitances of the generator, phase leads, lower part of the grounding resistor. As a result of the
and transformer low-voltage winding plus the ground- distributed capacitance of the generator and connected
ing system. Though the third harmonic voltage apparatus, a current flows and produces a voltage drop
changes with machine loading, the ratio of these values across the measuring transformer. A ground fault
remains relatively constant. occurring anywhere in the protected winding will cause
When a ground fault occurs, this relationship the current to rise and the voltage to increase across
changes and therefore allows detection of faults at all the grounding resistor. To differentiate from other
points not covered by the ‘‘neutral shift’’ relay. These conditions that could produce a similar voltage, the
two relays, 59N and 59D, provide ‘‘100% ground fault injection voltage is commutated, producing a coded
coverage.’’ Very high impedance faults in the protected character that is readily identifiable.
zone, of course, cannot be detected. Also for the unit- The GIX-104 may also be equipped to detect,
connected system neither relay will provide any backup through the injection principle, a ground fault in the
for faults on the transmission system. generator field.
Generator Protection 123

According to ANSI standards, the permissible

integrated product I22 t, (where I2 is negative sequence
current in per unit of machine rated current and t is
seconds duration) that ‘‘indirectly cooled’’ turbine
generators, synchronous condensers, and frequency-
changer sets can tolerate is 30. The standard for
hydraulic turbines or engine-driven generators is 40.
Standard ‘‘directly cooled’’ machines up to 800 MVA
are capable of withstanding a permissible integrated
product of 10, whereas some very large machines
(1600 MVA) can only tolerate 5. Early inspection to
detect damage is recommended for machines subject to
faults between the above limits and 200% of the limit.
Serious damage can be expected for faults above 200%.
With so many influences on the I22 t as shown in
Figure 8-8, it is evident that a relay designed to
respond in a way similar to the manner in which heat is
generated in the machine is mandatory. Many varia-
tions are available in generator negative sequence
overcurrent relays (see Fig. 8-9), but each is tailored to
match the I22 t ¼ K characteristic for K between 10 and
Figure 8-7 Neutral voltage injection. 40, with some spanning an even greater range. The
ANSI standard explicitly restricts this limit to I2 values
above the full-load level. Earlier relays covered this,
This scheme has the important advantage that it is but provided no I2 protection below the full-load
able to detect a stator or field circuit ground value. Later versions of the ANSI standard introduced
irrespective of the status of the generator. It may be an additional limit, unrelated to time. More recently
dead, spinning without field, near synchronizing, on- designed negative sequence overcurrent relays provide
line, loaded, or unloaded. an alarm level and trip capability for currents between
a pickup setting and full load. This is pertinent to
increased heating effects caused by such factors as
unbalanced load or faulty circuit breakers.
The negative sequence protective function is recom-
mended for all machines rated 1000 kVA or larger,
5.1 Unbalanced Faults though it can be justified for important smaller
machines. Examples of schematic connections for
Unsymmetrical faults produce more severe heating in unbalanced fault protection are shown in Figure 8-10.
machines than symmetrical faults. The negative The filter output that is applied across the operating
sequence currents that flow during these unbalanced coil of the COQ relay is
faults induce currents in the rotor having twice system
frequency. These currents tend to flow in the surface of VF a 2 K2 Ia2 ð8-1Þ
the solid rotor forging and the nonmagnetic rotor
when the connection shown in Figure 8-10a is used. If,
wedges and retaining rings. The resulting I22 R loss
however, the auxiliary current transformer is not used,
quickly raises the temperature. If the fault persists, the
metal will melt, damaging the rotor structure. VF a K2 ð2Ia2 þ Ia0 Þ ð8-2Þ
Such faults result from failure of a protective
scheme or equipment external to the machine. The where K2 is the filter constant.
relative magnitudes of negative sequence currents for If Ia0 is small, its effect can be ignored. Otherwise, it
line-to-line faults on a typical turbine generator under will be necessary to use either the auxiliary current
different operating conditions are shown in Figure 8-8. transformer to remove it or the relay with neutral
The effect of the higher excitation during the fault is made up inside, where its effect can be nullified. The
included for short circuits with load on the system. auxiliary current transformer is not normally required
124 Chapter 8

Figure 8-8 Relative magnitudes or negative sequence currents for line-to-line faults on a typical machine under different
operating conditions. (From AIEE Transactions, Volume 72, 1953, Part iii, Page 283, Figure 1.)

in unit-connected applications. For relays such as the Integrated protection relays such as the REG-100 or
SOQ and 46Q that use the equivalent of delta currents, REG-216 include the negative sequence protective and
any zero sequence current is ignored by the relay. alarm functions.
When a continuous load unbalance in excess of the
5 or 10% of the capability of the particular machine
may occur, the SOQ relay, or an additional relay 46Q, 5.2 Balanced Faults
set for the desired alarm level may be used to alert an
operator. With the SOQ, instrumentation can identify 5.2.1 Distance Relay (21)
the level of negative sequence current to permit a A generator should be protected also against damage
decision between tripping or decreasing the machine that will result from prolonged contribution to a
loading. balanced fault. A distance relay such as a KD-11, fed
An excellent backup for ground fault detecting from current transformers in the neutral of the
relays for a unit-connected generator utilizes a relay generator and a voltage supply connected at the
supplied by a current transformer (rated 100:5 or generator voltage level, provides such protection. A
thereabouts) in the secondary leads of the neutral single relay of this type complements the COQ, 46Q, or
distribution transformer. This is used by some utilities SOQ in recognizing balanced faults internal and
as the primary ground relaying. external to the generator. It also supplements these
Generator Protection 125

Figure 8-9 Comparison of relay and generator character- Figure 8-10 Schematic connections of the COQ relay for
istics. unbalanced fault protection.

5.2.2 Voltage-Controlled Overcurrent

relays by sensitively recognizing unbalanced faults.
Relay (51 V)
The connection described above makes the relay
directional from the neutral, but gives it a reach in If a negative sequence overcurrent relay is used, one 2-
both directions from the voltage transformer location. to 6-A 51-V relay also may be used to provide the
As a result, it will sense some generator as well as balanced fault backup function. A simple overcurrent
transformer faults. unit is unsuitable for preventing a sustained machine
The distance relay is usually set to reach through the contribution to a fault because, with a regulator out of
unit transformer. Unlike single-phase distance relays, service, the bolted sustained (or synchronous) three-
the reach of the KD-type relays is not affected by the phase fault contribution is less than machine full-load
phase shift through the bank. When set for an current. The 51-V relay, on the other hand, can be set
impedance greater than the transformer impedance, well below full-load current and not operate on load.
the KD-11 relay will operate for both generator and Its overcurrent unit is supervised or torque-controlled
line-side phase faults that manifest an impedance by an undervoltage unit, and therefore voltage must be
within this reach. A timer must be used to ensure below the voltage setting to permit the overcurrent unit
only the minimum equipment outage necessary to clear to function. Both the voltage and current units are
a fault. The timer must be set to coordinate with the independently adjustable, making coordination with
high-voltage transmission line relays and all other other overcurrent devices simpler than if the current
relays it overreaches. unit response were a function of voltage level.
126 Chapter 8

6 OVERLOAD PROTECTION Overexcitation can be caused by an attempt by a

generator voltage regulator to maintain rated voltage
6.1 RTD Schemes (Device 49)
during coast-down or holding manual excitation at a
fixed level during acceleration. Since the limits on
Most large generators are equipped with resistance
generators and transformers are inverse-time-related
temperature detectors (RTDs), which may be used in a
(i.e., a higher volts per hertz value is permitted for a
bridge circuit to provide sensing intelligence to an
shorter time to stay within the bounds of acceptable
indicator or a relay such as the DT-3 or 49 T.
heating), an inverse-time, volts-per-hertz relay such as
This relay is restrained when the resistance is low,
the MVH should be used to protect these devices when
indicating low machine temperature. When the tem-
overexcitation is likely. Figure 8-11 shows an example
perature of the machine exceeds some preset level such
of the coordination that can be achieved.
as 120 8C for class B insulated machines, the bridge
becomes unbalanced and the contacts close.

6.2 Thermal Replicas (Device 49)
A generator accelerates when it becomes separated
from its load. The acceleration depends on the inertia
A thermal replica relay utilizes stator current to
approximate the heating effects in the generator. The
machine thermal time constants on heating and cool-
ing are represented to take cognizance of previous and
present loading effects. When the replica indicates that
temperature in excess of the allowable value for the
machine insulation has been reached, tripping takes


From the fundamental expression for induced voltage

in a coil,
E ¼ 4:44 fANBm 108
E ¼ induced rms voltage in volts
f ¼ frequency in hertz
A ¼ cross-sectional area of the core in square inches
N ¼ number of turns
Bm ¼ flux density in maxwells per square inch
Since all the elements of the equation are constants
except E and f, it can be seen that
Bm a
Flux density is an excellent indicator of no load
heating effect. Hysteresis and eddy current losses are
each proportional to a power of the flux density.
Therefore, impending overheating can be recognized Figure 8-11 Example of volts/hertz protection using MVH
by measuring volts per hertz. relay.
Generator Protection 127

ðWR2 Þ, load loss, and governor response. To recognize

overspeed, a permanent magnet generator is often
connected to the machine shaft to supply a voltage to
the governor that is proportional to speed. The
governor may also be equipped with a speed-respon-
sive flyball mechanism. Either the permanent magnet
generator or flyball mechanism can initiate prime-
mover control to remove power input and alleviate
overspeed. An overfrequency relay, such as an MDF
(device 81), can be used to supplement this overspeed

9.1 Causes of Machine Loss of Field

Loss of excitation can occur as a result of

Loss of field to the main exciter
Accidental tripping of the field breaker
Short circuits in the field circuits
Poor brush contact in the exciter
Field circuit-breaker latch failure
Loss of ac supply to the excitation system Figure 8-12 Underexcited operation of generator.
Reduced-frequency operation when the regulator is
out of service
Some relays such as the KLF and KLF-1 contain
multiple operating units. The KLF and KLF-1 contain
(1) a directional unit, (2) an offset mho unit, and (3) an
and V to initiate partial shutdown if the operator is
instantaneous undervoltage unit.
unable to correct the problem quickly.
Loss of excitation produces voltage and current
These relays can also be used to detect loss of field
variations as shown in Figure 8-12. This causes a
in a synchronous condenser or motor. External faults
trajectory on the R-X diagram such as that depicted in
will cause the D and/or Z functions to restrain to avoid
Figure 8-13b. The initial operating point is dependent
undesirable tripping of the machine.
on load level and power factor angle prior to the loss of
Other relays such as the solid-state type 40 relay
excitation. When reactive power begins to flow into the
perform the function with only the reduced-diameter
machine, the locus moves into the X region. As the
impedance measurement.
transient continues, with the field collapsing, the locus
moves into the characteristic circle of the loss-of-field
relay. If the relay is equipped with an undervoltage
unit and at this time the voltage is sufficiently below
normal to operate it, tripping takes place after a short 9.2 Hazard
delay (X dropout). (See Fig. 8-13a.)
If reduction in terminal voltage is not appreciable, The generator must be kept on line, supplying power
the undervoltage unit does not drop out. This is as long as possible, particularly when the machine
indicative of a loss of excitation condition that is not represents a sizable portion of the system capacity. To
likely to affect system stability nor influence adjacent this end, an early warning of low excitation would give
machines significantly. For this series of conditions, it the operator an opportunity to restore the field if
will suffice to sound an alarm to alert the operator to possible and avoid tripping. Unnecessary tripping and
allow him or her to take action to restore the field or the resultant loss of kW output can precipitate system
anticipate shutdown. A timer is often run by Z and D breakup and a major outage.
128 Chapter 8

Figure 8-13 Trip circuits and R-X diagram showing operation of the KLF (40) loss-of-field relay.

9.3 Loss-of-Field Relays generator, supplying essentially the same kW to the

system as before the loss of excitation.
A relay designed to detect low excitation should Loss of synchronism does not require immediate
perform the following functions: tripping unless there is an accompanying decrease in
the terminal voltage that threatens the stability of
Alert the operator to any low excitation that could
nearby machines. Generally, it takes at least 2 to 6 sec
damage the machine or result in instability
to lose synchronism. Many instances have been
Alert the operator to a loss-of-field condition as
reported in which machines have run out of synchron-
early as possible, giving him or her time to
ism for several minutes because of loss of excitation
correct the condition
without damage to the machines.
Trip the machine automatically in the case of
impending system instability
The KLF-1 differs from the KLF relay in that it has
a separate phase voltage supply for each of three 9.4 KLF and KLF-1 Curves
different measuring elements. As a result, loss of any
one phase voltage to the KLF-1 relay cannot cause Figure 8-14 shows how a generator capability curve
incorrect tripping. With the KLF, some combinations can be transformed into an R-X diagram. For each
of load and phase voltage loss can operate the relay. point on the curve, an angle b can be measured from
The KLF-1 must have a wye-wye voltage supply, with the horizontal and the value of three-phase MVA read.
the neutral brought to the relay. The KLF may be used Knowing the line-to-line voltage at which the cap-
with a wye-wye, delta-delta, or open-delta-open-delta ability curve applies, a value of Z can be calculated
supply. using
When partial or complete loss of excitation occurs
on a synchronous machine, reactive power flows from kV2
Zp ¼ primary ohms ð8-3Þ
the system into the machine and the apparent MVA
impedance as viewed from the machine terminals kV2 ðRc Þ
(Vt/I) goes into the negative X region (Fig. 8-12). Z¼ secondary ohms ð8-4Þ
The kW output is controlled by the prime-mover input,
whereas kV Ar output is controlled by the field where Rc and Rv are the current and voltage
excitation. If the system is large enough to supply the transformer ratios, respectively.
deficiency in excitation through the armature, the Point Zp or Z can then be plotted, at angle b, on the
synchronous machine will operate as an induction R-X diagram. Other key points on the circle arcs can
Generator Protection 129

step of Figure 8-15a. To be useful in setting a loss-of-

field relay, these per unit values must be converted to
secondary ohms.
Figure 8-16 relates KLF (or KLF-1) setting to
capability and minimum excitation limiter (MEL)
curves. Assume a given kW load on the machine and
that the vars into the machine are gradually being
increased by decreasing machine field current,
producing a trajectory, as in curve A of Figure 8-
16. If the regulator is in service, the MEL prevents
operation at a level that would jeopardize the
machine thermally. If the regulator is out of service,
Z continues to decrease until the KLF impedance
unit operates. An alarm indicates a hazardous
operating condition if the voltage is high. A low
voltage, which may seriously jeopardize system
stability, trips the machine after 0.25 sec (Fig. 8-
13a). The loss-of-field relay must reach into the plus
X area if its locus is to follow closely the machine
characteristic. A directional unit is included in the
relay to avoid tripping for close-in faults beyond the
unit transformer.

9.5 Two-Zone KLF Scheme

Like all other elements in generator protection, the

loss-of-field relay should be supported by backup to
Figure 8-14 Transformation from KW-KVAR plot to R-X
prevent catastrophic failure if a device is out of service
or a component should fail.
Two loss-of-field relays provide better protection
than one. The first, or zone 1, relay is set to be
restrictive (Fig. 8-17) and typically trips through a
0.25-sec timer. It provides fast clearing on loss of field,
be obtained in the same way until the entire capability yet is secure against swings such as that shown passing
curve is transformed. through points CDEF in Figure 8-17. The zone 2 relay
The steady-state stability curve is another signifi- is set wider and typically drives a 1-sec timer to detect
cant limit that can be related to a loss-of-field relay partial loss of field, provide an alarm function, and
with impedance-measuring qualities. The MW- back up the zone 1 relay. A total functional
MVAR curve can be developed as shown in characteristic similar to the KLF type is required to
Figure 8-15a. In this figure, V is the per unit terminal do this. Other setting data are given in Tables 8-1 and
voltage, Xs the equivalent per unit system impedance 8-2.
as viewed from the generator terminals, and Xd the Figure 8-18 shows the dc schematic for the KLF or
per unit unsaturated synchronous reactance of the KLF-1 zone 2 relay. Zone 1 may be used without a
generator. Both Xs and Xd are measured on the TD-A timer unless extreme emphasis on security is
machine MVA base. made. No TD-2 relay is required if the undervoltage
Figure 8-15b converts the machine’s steady-state contacts of the zone 1 relay are shorted or the type 40,
stability curve to an R-X diagram. Note that the curve solid-state relay is used. Without TD-2 quick tripping
of Figure 8-15b can be plotted directly from a occurs when Z and D operate irrespective of the
knowledge of Xs and Xd without the intermediate voltage level.
130 Chapter 8

Figure 8-15 Conversion of steady-state stability curve to R-X.

10 PROTECTION AGAINST GENERATOR Due to the extreme nature of this hazard, two
MOTORING motoring relays are recommended to operate in an
OR mode and respond to different current and
Generator motoring (sometimes referred to as reverse voltage. The reverse-power relay is commonly used
active power) protection is designed for the prime with diesel engine generating units, particularly when
mover rather than the generator. With steam the danger of explosion and fire from unburned fuel
turbines, for example, turbines will overheat on low exists.
steam flow, but will be protected by steam tempera- Motoring results from a low prime-mover input to
ture devices. With hydroturbines, hydraulic flow the ac generator. When this input cannot satisfy all the
indicators protect against blade cavitation on low losses, the deficiency is supplied by the generator
water flow. Similar devices are used to protect gas absorbing real power from the system. Since field
turbines. excitation should remain the same, the same reactive
Generator motoring protection can be provided by power would flow as before motoring. Thus, on
devices such as limit switches or exhaust-hood motoring, the real power will flow into the machine,
temperature detectors. However, a reverse-power whereas the reactive power may be either flowing out
protective relay is recommended for added safety. of or into the machine. Usually, the reactive power will
Generator Protection 131

turbine had its valves closed to slightly less than the

no-load steam requirements. The turbine would
supply, say, 99% of the losses, and the generator (as
a motor) would supply 1%. If the total losses were
3.0% of the kW rating, the kW drawn by the generator
(as a motor) from the power system would be only
1.0% of 3.0%, or 0.03%, of the nameplate rating. This
is a challenge beyond the capabilities of most motoring
detection relays. Solid-state and microprocessor tech-
nology allows such low sensitivity while, at the other
end of the scale, providing full-load current capability.
The SRW relay may be set as low as 1 mA.
When the prime mover is spun at synchronous
speed with no power input, the approximate reverse
power required to motor a generator, as a percentage
of the nameplate rating in kW, is as follows:
Condensing steam turbine 1 to 3%
Noncondensing steam turbine 3þ%
Figure 8-16 KLF setting related to capability and MEL Diesel engine 25%
Hydraulic turbine 0.2 to 2þ%

be supplied to the system as machines are not generally 10.1 Steam Turbines
operated underexcited.
A relay designed to detect motoring must be When operating under full vacuum and zero steam
extremely sensitive and even then cannot detect all input, condensing turbines require about 3.0% of the
conditions of reverse power. For example, suppose a kW rating to motor. Noncondensing turbines require
3.0% or more of the rated kW to motor when
operating against atmospheric or higher exhaust
pressures at zero steam flow.

10.2 Diesel Engines

If no cylinders are firing, diesel engines require about

25% of the rated kW figure. If one or more cylinders
are firing at no load, the reverse power will be lower
than this depending on the governor action and effect
on the system frequency.

10.3 Gas Turbines

The large compressor load of gas turbines represents a

substantial power requirement from the system when
motoring. Consequently, the sensitivity of the anti-
motoring device is not critical.

10.4 Hydraulic Turbines

Figure 8-17 KLF relay stable swing following clearing of When the blades are under the tail-race water level,
nearby three-phase fault. the percent of kW rating required for motoring is
132 Chapter 8

Table 8-1 Recommended Settings for KLF Relay

Setting Zone 1 (alone) Zone 2 (alone) Both zone 1 and zone 2

Impedance setting See Figure 8-17. See Figure 8-17. See Figure 8-17.
Voltage setting (a) Undervoltage contact 80% Zone 1 voltage contact
shorted or shorted.
(b) set at 80% for security. Zone 2 dropout voltage set at
TD-1 (see Fig. 8-18) 1/4 to 1 sec (1/4 see adequate). 1/4 to 1 sec (1 sec preferred). Zone 1 timer ¼ 1/4 sec.
Zone 2 timer ¼ 1 sec.
TD-2 (see Fig. 8-18) Not required for (a) above. 1 min. None for zone 1. zone 2
For (b) above use 1 min. timer ¼ 1 min.
Advantages Less sensitive to stable system (1) More sensitive to LOF (1) Same as (1), (2), and (3) at
swings. condition. left.
(2) Can operate on partial (2) Provides backup
LOF. protection.
(3) Provide alarm features for
manual operation.

probably well over 2.0%. From 0.2 to 2.0% kW is 11 INADVERTENT ENERGIZATION

required for the turbine to motor when the blades
are above the tail-race level. For turbines using a Many instances of inadvertent energization have
Kaplan adjustable-blade propeller, the flat-blade occurred over the years. A few of the causes were
condition probably requires less than 0.2% kW to
Closing the generator breaker with the machine at
Most commonly, a single-phase relay is used for
Closing a station service breaker with the machine
motoring detection, with sensitivities ranging from
at standstill
1 mA to 5 A. All have adjustable operating times.
High-voltage breaker flashover near synchronism
When the more sensitive relays are used, care should be
Closing of generator disconnect with unit breaker
used in selecting a current transformer for them.
Indeed, a metering accuracy-class ct is preferred to a
relaying accuracy class. A typical schematic is shown in Although careful operating procedures and the judi-
Figure 8-19. cious use of interlocks can usually prevent these

Table 8-2 Special Settings for Multi-machines Bussed at Machine Terminals

Setting Zone 1 (alone) Zone 2 (alone) Both zone 1 and zone 2

Impedance setting See Figure 8-17. See Figure 8-17. See Figure 8-17.
Voltage setting (a) Undervoltage contact 87%. Zone 1 voltage contact
shorted or shorted with zone 2 set at
(b) set at 87% for security. 87%.
TD-1 (see Fig. 8-18) 1/4 to 1 sec (1/4 sec adequate). 1/4 to 1 sec (1 sec preferred). Zone 1 timer ¼ 1/4 sec.
Zone 2 timer ¼ 1 sec.
TD-2 (see Fig. 8-18) Not required for (a) above. 10 sec for directly cooled. None for zone 1.
For (b) above use 10 sec for 25 sec for indirectly cooled. Zone 2 timers: 10 sec for
directly cooled, 25 see for directly cooled, 25 sec for
indirectly cooled. indirectly cooled.
Generator Protection 133

Figure 8-18 Type KLF or KLF-1 dc schematic for zone 2 loss of excitation protection. (Timer settings are given in Tables 8.1
and 8.2.)

occurrences, the ingenuity of humans to circumvent may operate, but again the time delay for tripping may
these procedures and interlocks is legendary. be unsuitable.
Full-voltage energization of a machine at standstill To further compound the difficulties associated with
does not produce an enormous magnitude of current, detecting inadvertent energization is the fact that
but it does supply an extreme impact of torque, and generator potential circuits are often disconnected in
mechanical damage to the shaft or bearings may occur. the interests of safety when a machine is shut down.
The resulting current is of sufficient magnitude that Any of the ‘‘normal’’ relays that are dependent on this
fast removal is necessary if thermal damage to the voltage supply will be unable to respond at the very
generator is to be avoided. Instantaneous separation time when they are needed. Also, it must be
offers no guarantee that no damage will occur. remembered that the flashover of an open breaker
Various relays applied for other functions may cannot be cleared by energizing its trip coil.
detect inadvertent energization. Loss-of-field relays Several protective schemes have been used success-
may respond, but they usually use single-phase current, fully to detect three- or single-phase inadvertent
and therefore, complete sensing is not afforded for all energization. Among them are
possible combinations of this phenomenon. Relays
applied to detect motoring will operate, but their Directional overcurrent relays
operating time is set to other criteria and the long time Pole disagreement relays
customarily used is unsuitable for this additional Relays containing logic to detect overcurrent for a
function. Distance- and voltage-controlled overcurrent short time following 0 V
relays applied for generator phase backup protection Frequency-supervised overcurrent relays
134 Chapter 8

Figure 8-19 Typical schematic for antimonitoring protection using CRN-1 (32) relay.

Voltage-supervised overcurrent relays 12 FIELD GROUND DETECTION

Distance relays
Unit transformer neutral overcurrent in special A single ground on the field of a synchronous machine
breaker-failure relaying scheme. produces no immediate damaging effect. It must be
detected and removed because of the possibility of a
Care must be exercised to assure that either a voltage second ground that could short part of the field
supply is available when operation is required or winding and cause damaging vibration. Care should be
undervoltage allows operation of the scheme. Further, used in establishing any field ground detecting scheme
the circumstances associated with inadvertent energi- to assure that any bearing current that is allowed will
zation itself must not be able to circumvent a necessary not cause bearing deterioration. One scheme that has
part of the logic for tripping, such as the requirement been used is shown in Figure 8-20. A small leakage
for the presence of reduced frequency. current flows through the field-to-ground capacitance,
Generator Protection 135

which, on a large turbogenerator, can be between 0.3 The magnitude of this voltage will vary according to
and 0.5 mf. The relay detects an increase or decrease in the exciter voltage and point at which the field is
the magnitude of this current. grounded. The voltage will be at maximum if the field
is grounded at either end of the winding.
A null point will exist in the field winding where a
12.1 Brush-Type Machine ground will produce no voltage between M and
ground. This null point will be located at a point on
One recommended field ground protection scheme for the field winding from which there is balance between
a generator with brushes (i.e., stationary field leads the two field winding resistances and two relay
accessible) is illustrated in Figure 8-21. This scheme, resistances to positive and negative. A ground at the
which does not require any external source, uses the null point will go unrecognized until the field voltage is
very sensitive d’Arsonval dc relay, type DGF. changed as a result of daily reactive (or voltage-level)
The DGF relay uses a voltage divider circuit, scheduling. Faults having impedance of up to
consisting of two linear resistors (R1 and R2) and a 300,000 O can be recognized with this scheme.
nonlinear resistor whose resistance varies with the A pushbutton, connected across a portion of the R2
applied voltage. If the field becomes grounded, a resistor, permits a manual check for possible ground
voltage will develop between point ‘‘M’’ and ground. faults at the center of the winding. This provision is

Figure 8-20 Path of the currents in a machine when using an ac field ground relay.
136 Chapter 8

This requires a pilot brush connected to the neutral of

the exciter that is periodically dropped.
Another variation of this form of detection is used
by the solid-state type 64F relay. It impresses a dc
voltage from the negative dc lead to ground and
continuously monitors current flow. An increase in
current accompanies a field ground.
The pilot brush arrangement can also be used with
the Wheatstone bridge scheme (YWX111) described in
Section 12.1.

12.3 Injection Scheme for Field Ground


The scheme described in Section 4.4, GIX-104, is

equally applicable to field ground detection, irrespec-
tive of the type of excitation system in use. This is
shown in Figure 8-22 for a ‘‘rotating rectifier’’
excitation system. As with the stator winding ground
detection application, this system is able to detect field
circuit grounds even though the machine is at standstill
or running, excited or not.

Figure 8-21 Field ground protection scheme for a gen- GENERATORS
Alternating-current overvoltage protection is recom-
mended for hydroelectric generators subject to over-
speed and consequent overvoltage on loss of load.
desirable when the generator is to be ‘‘base-loaded’’ Some hydroelectric generators can go up to 140% or
and will not experience periodic excitation variations. more of rated speed when full load is dropped. The
Another scheme that has been used successfully voltage may reach 200% or more.
utilizes a Wheatstone bridge with the field circuit to The ac overvoltage protective scheme is shown in
ground forming one leg of the bridge. A solid ground Figure 8-23. The relay, which changes the excitation to
fault anywhere in the field circuit can be detected reduce the output voltage, can also provide backup
immediately through the recognition of the resulting protection for the voltage regulator.


12.2 Brushless Machines FREQUENCIES

For a ‘‘brushless’’ type of machine, no normal access is Many turbine generators are started on turning gear,
available to a stationary part of the generator field which rotates the shaft at about 3 rpm. For a cross-
circuit, and no continuous monitoring to detect field compound machine, the field must be applied before
grounds is possible. One widely employed scheme uses the machine is removed from the turning gear. At this
a 60-Hz tuned overvoltage relay connected between the point, excitation should be limited to rated volts per
neutral of the three-phase ac exciter and ground. A hertz to avoid overexciting the unit or station service
ground on the exciter, in the three-phase rectifier transformer. A tandem unit need not have field applied
bridge, in the field, or on the dc leads will be detected. until it is up to speed and ready to synchronize.
Generator Protection 137

Cross-compound generators may be operated for

several hours during warmup at frequencies well below
their rating. Current transformer and relay perfor-
mance must be considered at these reduced frequencies
because fault magnitudes are approximately the same
as at rated frequency. Current transformer perfor-
mance can be expected to deteriorate badly at low
frequency. There may be a small compensating effect,
however, in the reduction of burden impedance.
The performances of some electromechanical relays
associated with the generator or a generator-transfor-
mer unit at 15 and 30 Hz are summarized in Table 8-3
in terms of 60-Hz performance.
For cross-compound turbine generators, the low-
and high-pressure units should have their fields applied
and be synchronized while on turning gear, so that
they are brought together up to rated speed. Synchro-
nizing surges may occur, at these low speeds, that will
operate the loss of field relays. These synchronizing
surges can reach 60% of the full-load current, since the
impedance in the generators is very low. Loss-of-field
relays applied to a cross-compound unit should there-
fore be disabled during startup.
The SC relay (and/or the SV) is recommended
as supplementary protection when reduced fre-
quency protection is required. The SC current
relay has a flat characteristic and increases slightly
in sensitivity as the operating frequency drops.
When an SC relay is operated on dc, it picks up
at approximately 15% below its normal 60-Hz
Figure 8-22 Rotating rectifier excitation system.
pickup. The pickup of the SV voltage relay is
almost directly proportional to frequency; its
sensitivity at 15 Hz is thus 4 times the sensitivity
at 60 Hz. For this reason, the SV relay provides
excellent reduced-frequency protection.
In this application, an SC relay is frequently located
in the differential circuit of each phase of the generator
differential relay. The SV relay is connected across the
secondary resistor in the generator neutral circuit.
None of the relays listed in Table 8-3 will overheat
nor operate incorrectly if left in the circuits when the
generator is operated at reduced frequencies. The KLF
and KLF-1 relays, when used in a cross-compound
configuration, must have their trip incapacitated
during startup. The SC and SV relays used for low-
frequency protection must be removed from service for
normal operation. This may be done by a frequency
relay set for approximately 55 Hz (for a 60-Hz system)
and an auxiliary relay.
The microprocessor-based multifunction relay Type
GPU2000R has frequency-tracking algorithms. These
Figure 8-23 Overvoltage protection for generator. insure that its protective elements which are most
138 Chapter 8

Table 8-3 Performance of ABB Protective Relays at Reduced Frequencies

Pickup in percent of 60-Hz pickup

15 Hertz 30 Hertz Classificationa

Overcurrent (51) CO-2 165 115 A

CO-5 150 B
CO-6 143 B
CO-7 140 B
CO-8 262 138 A
CO-9 260 140 A
SC 85 93 A
COV Performance same as CO Unit used in relay
Voltage (59) SV 26 50 A
CV contact-making 122 120 A
b b
CV-8 C
Differential (87) CA generator 255 123 A
CA transformer 149 B
SA-1 370 175 A
HU-1 250 130 A
Negative sequence COQ O
(46) SOQ
Loss of field (40) KLF (or KLF-1) D
(A) Protection available at both 15 and 30 cycles. (B) Protection available at 30 cycles only. (C) Additional protective relays required for start-up
or low-frequency operation. (O) The sensitivity of the COQ & SOQ relays to negative sequence currents is a direct function of frequency, while its
sensitivity to positive sequence currents is an inverse function of frequency. This relay will operate for heavy three-phase and phase-to-phase faults
at reduced frequencies, but should not be relied upon for primary protection during warm-up. (D) Since the KLF or KLF-1 relays operate on
lagging reactive power into the machine, the relay will neither operate falsely nor provide loss-of-field protection during the warm-up period.
Very insensitive or nonoperable at the frequency indicated.

needed during low-frequency operation maintain their two categories: (1) those having the longest stage
characteristics. blading of length 18 to 25 in. and (2) those having the
longest stage blading of length 28 12 to 44 in. These
curves are not a national or international standard.
15 OFF-FREQUENCY OPERATION The specific manufacturer of the turbine should be
contacted to ascertain specific recommendations for
Turbine blades are carefully designed to have no limits.
mechanical resonant conditions when rotating at rated Under- (or over-) frequency relays as described in
speed. If this were not the case, mechanical deteriora- the chapter on ‘‘load-shedding’’ may be used to detect
tion would occur as the blades flex under the stress of frequency excursions, but this must be monitored by
loaded operation. At elevated or reduced speed, there watt level because off-frequency operation at a light
are resonant points where prolonged operation pro- load (above the level that produces adequate steam
duces blade fatigue damage and ultimate failure. flow to remove turbine friction and windage losses) is
If operating frequency (speed) deviates from the not hazardous. The combination of these two sensing
rated value, corrective action must be initiated or elements is used to drive an accumulation timer to
tripping must result. Since mechanical fatigue is a allow an estimate of the extent of life remaining to be
cumulative phenomenon, the time of loaded operation made. The need for blade examination and possible
at reduced frequency must be monitored and accumu- replacement can then be evaluated.
lated over the life of the turbine. Figure 8-24 shows the The microprocessor-based system REG-216 has a
limits that one manufacturer imposes for machines in provision for integrating low-frequency overtime and
Generator Protection 139

Figure 8-24 One manufacturer’s limits for off-frequency operation of combustion turbines.

alarming and tripping to prevent damage to the described in Chapter 14, ‘‘System Stability and Out-
turbine. Also, to avoid any detrimental effects of of-Step Relaying.’’
the front-end filters in detecting overcurrent at off-
frequency levels, the 50, instantaneous-trip function
is sensed ahead of these filters. Overcurrent protec-
tion, then, is provided at a frequency as low as
2 Hz.
The automatic transfer of highly essential station
auxiliary loads such as reactor coolant pumps, boiler
16 RECOMMENDED PROTECTION feed pumps, and induced draft fans is common
practice. Paralleling the normal and emergency sources
Figure 8-25 shows the recommended protection for on a continuous basis is not generally recommended
large tandem-compound unit-connected turbine gen- because the higher breaker-interrupting duties
erators. Figures 8-26 and 8-27 illustrate the recom- involved can cause problems, as can circulating
mended protection for machines that are not unit- currents between systems. Transfers should not be
connected. Generally, such generators are used in made if voltage of the alternate supply is not
industrial applications. satisfactory or the load circuits are faulted. Also,
A wide variety of implementations of the numbered supply breaker tripping should be delayed long enough
functions are in use and some are described in Table to permit fault sectionalizing in the load circuits.
8-4. The REG-100 and REG-216 are complete multi-
function microprocessor-based packages.
18.1 Fast Transfer

This is a term applied to the connection of a bus to a

17 OUT-OF-STEP PROTECTION second power source with little or no time delay. This
process is accomplished with an ‘‘open’’ transfer or
As generator impedances become larger in proportion ‘‘closed’’ transfer. The open transfer disconnects the
to the system impedance, the electrical center will be normal source before connecting the second. The
closer to the generator. This condition intensifies the closed transfer allows the two systems to operate tied
need for out-of-step detection as part of the generator together momentarily, and then to have the original
relaying complement. Such relaying schemes are source breaker tripped. For all the fast transfer
140 Chapter 8

Figure 8-25 Overall protection for a tandem-compound unit-connected generator.

schemes, it is recommended, if a synchronism check The simple process of tripping breaker A, when in
relay is required, that it be preenergized so that it may the automatic mode in the absence of a bus fault,
determine the existence (or lack of) synchronism prior produces the closure of breaker C.
to the need to transfer.
Figure 8-28a shows an example of the fast open-
transfer scheme. 43 is a switch having automatic and
manual positions. With 43A closed, CVX synchronism 18.2 Choice of Fast Transfer Scheme
check relay contact closed due to the two buses being
in synchronism prior to the need for transfer, breaker Open transfer would be selected in a system in which the
A having been tripped for any reason (as evidenced by supportive WR2 (moment of inertia) is sufficient to keep
its 52b being closed), and no fault having occurred on the switched bus close enough in frequency during the
the bus to produce the opening of 86B, then the close open period to allow reconnection without an excessive
circuit of breaker C becomes energized. shock to rotating machinery connected to that bus.
Generator Protection 141

Figure 8-26 Recommended protection for large generators as used in industrial plants.

Figure 8-27 Recommended protection for small generators as used in industrial plants.
142 Chapter 8

Table 8-4 Generator Relaying

device no. Description Traditional Solid-state or numerical GPU-2000R REG 216 REG 316

2 Timer — 62T X X X
21 Phase backup KD-11 X X X
24 Volts/hertz — 59F X X X
32 Motoring CRN-1 32R X X X
40 Loss of field KLF 40 X X X
46 Negative sequence COQ 46Q X X X
49 Thermal DT-3 49T — X X
50/51 Stator overcurrent CO-ITH Micro 51 X X X
59 Overvoltage CV-5 59 X X X
59F Field ground CV-8 59G — X X
59N Stator ground (95%) CV-8 59G X X X
64S Current injection — GIX-104 — X
64R (100% ground stator/rotor)
67 Inadvertent-energization CRG-9 32D X X X
76 Field dc overcurrent D-3 76H — — —
78 Out-of-step KST-KD-3 GZX-104 — X X
81 Underfrequency — 81 X X X
86G Generator lockout LOR LOR LOR LOR LOR
87G Generator differential CA 87M X X X
87N3 Stator-neutral-ground DGSH 27G X — X
87T Overall differential HU-1 87T — X X
Alternate 51L þ 47H or 51V function.

The ‘‘simultaneous’’ scheme is a variation of the Synchronism check is a satisfactory means for
open transfer scheme. With it, the trip-coil of the supervising the closing of the bus tie breaker. It is
normal supply breaker and the close-coil of the not suitable, however, for a fault-forced transfer unless
incoming supply breaker are energized simultaneously. the system is supervised by an 86B lockout relay that
Since the close function is somewhat slower than the prevents transferring to a bus fault. Otherwise, it
trip function of a given type of breaker, the opening would be possible for both sources to be connected to a
breaker wins the race. Figure 8-28c shows the contact fault, compounding the damage and producing no
array for this sequence. useful result.
Closed transfer would be the proper scheme to use
when WR2 associated with the rotating equipment is
inadequate to allow separation for even a short time.
The impact of reenergization would be too great.
Figure 8-28c shows the addition of a ‘‘52a’’ contact of 18.3 Slow Transfer
breaker C to delay the tripping of breaker A on a
manual transfer. It is not considered wise to delay Slow transfer carries with it the connotation of
tripping for a fault that operates 86S, the lockout relay delaying the reenergization of a bus until the voltage
associated with a source circuit fault. on motors connected to the bus has decayed to the
Generator Protection 143

point where no damage can be expected with out-of-

phase energization. This level is generally considered to
be 25% of rated. Figure 8-28b describes this arrange-


Important features related to generator protection can

only be achieved by a coordinated generator protection
package, utilizing microprocessors. Self-monitoring,
communications, oscillography, and adaptive settings
are accomplished straightforwardly and reliably. Flex-
ibility in the application is important to allow
individual user selections. This is accomplished by a
man-machine interface that allows ease of settings and
in some cases the choice of software modules.
In the use of multifunction coordinated packages,
sight should not be lost of the need for adequate
redundancy to provide backup and cover the failure of
any element. The REG 216 and REG 100 series
address these needs and provide extensive generator
protection utilizing proven concepts.

Figure 8-28 Variations of transfer schemes.

Motor Protection
Revised by: C. L. DOWNS

1 INTRODUCTION 4. Phase reversal. Starting a motor in reverse can

be hazardous to the load.
1.1 General Requirements
5. Phase unbalance. A small amount of unbalance
can result in a significant increase in the motor
Motor protection is far less standardized than gen-
erator protection. Although the National Electric Code
6. Out-of-step operation for synchronous
and NEMA standards specify basic protection require-
ments, they do not fully cover the many different types
7. Loss of excitation for synchronous motors.
and sizes of motors and their varied applications. There
are many other schemes, all of which offer different
Protective relays applied for one hazard may operate
degrees of protection. As with generator protection, the
for others. For example, a relay designed to operate on
cost and extent of the protective system must be
an excessive overload could also protect against a fault
weighed against the potential hazards. The size of the
in the windings.
motor and type of service will also influence the type of
Protective devices may be installed on the motor
protection required. Electromechanical, solid-state, or
controllers or directly on the motors. The protection is
microprocessor-based relays can be used stand-alone or
usually included as part of the controller, except for
in combination with one another to achieve the desired
very small motors, which have various types of built-in
degree of security and dependability.
thermal protection.
Motor protection should involve the detection of
Motors rated 600 V or below are generally
the following hazards:
switched by contactors and protected by fuses or
1. Faults in the windings or associated feeder low-voltage circuit breakers equipped with magnetic
circuits, including both phase and ground fault trips. Motors rated from 600 to 4800 V are usually
detection. switched by a power circuit breaker or contactor
2. Excessive overloads. Overloads result in ther- (often supplemented by current-limiting fuses to
mal damage to the insulation and can be caused accommodate higher interrupting requirements.
by continuous or intermittent overload, or a Motors rated from 2400 to 13,800 V are switched
locked rotor condition (failure to start or a jam by power circuit breakers.
condition). Although protective relays may be applied to a
3. Reduction or loss of supply voltage. Any re- motor of any size or voltage rating, in practice they are
duction of supply voltage directly affects the usually applied only to the larger or higher-voltage
applied torque to the connected mechanical load. motors.

146 Chapter 9

1.2 Induction Motor Equivalent Circuit

Excessive heat in the motor can be caused during

starting, a locked rotor condition, load requirements,
voltage unbalance, or an open-phase condition and
can cause degradation of the mechanical and dielectric
strength of the insulation. This thermal deterioration
of the insulation sets up the possibility of future faults.
The equivalent circuit of the motor helps in
visualizing what occurs in the motor during the above
conditions. The motor impedance during these condi-
Figure 9-2 Induction motor negative sequence network.
tions is directly influenced by the slip of the motor:
ns  nr the starting current Is of the motor is 8IFL, then I2
slip s ¼ per unit would be 0.40, or 40%. This will result in an increase in
stator heating and, in particular, rotor heating. The
where ns and nr are the stator field and rotor speeds. At rotor heating results from the combined effect of the
start, the slip is 1.00, or 100%. During a running counterrotating flux that causes large currents to be
condition, the slip is approximately 0.01 to 0.08, or 1 to induced at fs(2  s) Hz, where fs is the system
8%. The equivalent circuit during a starting condition frequency, and the increased skin effect in the rotor
is as shown in Figure 9-1. The input impedance at start that can cause its resistance to be 5 to 10 times normal.
can be approximated by

Zstart ¼ Rs þ Rr þ jXs þ jXr 1.3 Motor Thermal Capability Curves

where Rs, Rr, jXs, jXr are the stator and rotor
Protecting a motor for a variety of hazards requires the
resistances and reactances. During a start condition,
protection engineer to know full-load current, permis-
the impedance is predominately reactive. As the motor
sible continuous allowable temperature rise, locked-
gains speed, the impedance becomes more resistive,
rotor current and permissible maximum time at that
and the power factor increases.
current, and accelerating time, which is a function of
The negative sequence impedance of a motor,
the load characteristics and starting voltage. A typical
excluding wound rotor motors, is very nearly equal
motor thermal capability curve, which is shown in
to Zstart above, and from Figure 9-2, can be
Figure 9-3, is helpful in determining the temperature
approximated by
endurance of the insulation. The lower part of the
RR curve is usually rotor-limited. The limit arises because
Z2 ¼ Rs þ þ jXs þ jXR of the I2R heating effect during a locked rotor
With Z2 and Zstart approximately equal, the negative
sequence current can be calculated for a particular
negative sequence voltage unbalance. If V2 ¼ 5% and

Figure 9-1 Induction motor equivalent circuit at start. Figure 9-3 Motor thermal capability curve.
Motor Protection 147

condition. This represents the time a motor can remain

stalled after being energized before thermal damage
occurs in the rotor. This is an I2t limit, which can also
be expressed as a (V2/R)t limit. The middle portion is
the acceleration thermal limit part of the curve. This is
from the locked-rotor current to the motor breakdown
torque current portion of the curve. The upper section
of the curve is the running or operating thermal limit
portion. This represents the motor overload capacity.


The phase-fault current at the terminals of a motor

usually is considerably larger than any normal current,
such as starting current or the motor contribution to a
fault. For this reason, a high-set instantaneous-trip
unit is recommended for fast, reliable, inexpensive,
simple protection. When the starting current value
approaches the fault current, however, some form of
differential relaying becomes necessary. The sensitivity
of the differential relay is independent of starting
current, whereas instantaneous-trip units, which Figure 9-4 Comparison of sensitivities of type CA differ-
respond to phase current, must be set above the ential relay and IT instantaneous trip unit.
starting current (including any dc offset due to
asymmetrical transients that may be caused by voltage
switching). This difference is shown in Figure 9-4. The maximum motor starting current in this case is
To allow for fault resistance and different types of 1
faults and to assure twice pickup on the unit for ILR ¼
ð0:08 þ XM Þ
minimum fault, the instantaneous phase-relay pickup
should be set at less than one-third of I3ph, where I3ph is where XM is the motor impedance. In order that
the system contribution, excluding the motor contribu- I3ph/ILR > 5, XM must be greater than 0.32 per unit on
tion, to a symmetrical three-phase fault on the motor the transformer-rated KVA base.
feeder. Also, pickup should be set at 1.6 times ILR or If the motor has a full-voltage starting current of six
more, where ILR is the actual symmetrical starting times full load, then XM ¼ 1/6 ¼ 0.167 on the motor-
current, as limited by source impedance. The ratio I3ph/ rated KVA base. With a motor KVA of one-half the
ILR should thus be greater than approximately 5.0. transformer KVA, an XM of 0.167 would be 0.333 on
In general, then, instantaneous-trip units can be the transformer base, and greater than 0.32. Clearly,
used for phase protection if the motor KVA (or this rule of thumb should only be applied when there is
approximately the horsepower) is less than one-half no appreciable deviation from the parameters assumed
the supply transformer KVA. If not, differential above.
protection, such as that obtained by the CA or 87M
(Fig. 9-4), is required for sensitive fault detection.
The logic for this criterion comes from the follow- 3 GROUND-FAULT PROTECTION
ing. Assume a motor is connected to a supply
transformer with 8% impedance. The maximum fault A solidly grounded system may be protected by an
current at the transformer secondary with an infinite inverse, very inverse, or short-time induction or
source is microprocessor-based relay connected in the current
transformer residual circuit. For a solid fault at the
I3ph ¼ 1=0:08
machine terminals, a typical setting is one-fifth of the
¼ 12:5 per unit on the transformer base minimum fault current. Time dial settings of around 1
148 Chapter 9

give operations of four to five cycles at 500% pickup

when the CO-2 relay is used.
During across-the-line starting of large motors, care
must be taken to prevent the high in-rush current from
operating the ground relays. Unequal saturation of the
current transformers produces a false residual current
in the secondary or relay circuits. Using two- rather
than three-phase relays or three-phase relays with
different impedances will tend to increase the effects of
false residual currents.
False relay operation is unlikely if the phase Figure 9-5 BYZ ground relaying scheme.
burdens are limited so that the voltage developed by
the current transformer during starting is less than 75%
of the relaying accuracy voltage rating of the current winding only when zero sequence current is flowing in
transformer for the particular CT tap being used. If the primary leads. Since virtually all motors have their
false relay operation is a problem, the ground relay neutrals ungrounded, no zero sequence current can
burden of an electromechanical relay should be flow in the motor leads unless there is a ground fault
increased by using a lower tap. All three transformers on the load side of the BYZ. If surge-protective
will then be forced to saturate more uniformly, equipment is connected at the motor terminals,
effectively reducing the false residual current. This however, current may be conducted to earth by this
increased saturation may reduce the sensitivity to equipment. To date, there has been no reported case of
legitimate ground faults and this should be checked. an instantaneous relay connected to a BYZ current
Alternatively, a resistor or reactor can be connected in transformer tripping because of surge-protective
series with the ground relay. equipment. The presence of such equipment may safely
The common practice in 2400- to 14,400-V station be ignored in choosing a relay.
service, and industrial power systems, is to use low- Solid-state relays such as the type 50D when used
resistance grounding. By using the ‘‘doughnut current with the BYZ give good performance in this applica-
transformer’’ scheme, such systems offer all the tion due to a lower burden characteristic. Also,
advantages of instantaneous trip units—speed, relia- specialized solid-state ground-fault relay systems such
bility, simplicity, low cost—without any concern for as the Ground-Shield2 series provide a variety of
starting current, fault contributions by the motor, false doughnut CT window sizes, both toroidal and
residual current, or high sensitivity. rectangular. As a system, the relay and CT character-
Figure 9-5 shows how the BYZ zero-sequence-type istics are properly prematched by design, and thus need
current transformer can be used as a supply for the 50 not be further considered by the user. These specialized
instantaneous-trip (IT) unit or 51 time overcurrent systems usually have relay pickup settings marked in
(CO) relay. Typical sensitivities obtainable with these terms of primary amperes.
ground-fault protection systems are shown in The BYZ current transformer is also used in the flux
Table 9-1. A voltage is generated in the secondary balancing differential scheme, in which each phase is

Table 9-1 Relay Settings and Sensitivities Using the 50/5 BYZ Zero Sequence Current Transformers

Minimum sensitivity in primary 310 Maximum primary 310 amperes for

Relay type Relay setting amperes accurate timing and coordination

43=4 IDa 73=4 IDa 43=4 IDa 73=4 IDa

IT 0.15 5.0 5.0 — —
CO-8 or 9 0.5 9.0 10.0 25 112
CO-8 or 9 2.5 24.0 24.0 540 1215
CO-11 0.5 6.0 7.0 70 150
CO-11 2.5 24.0 24.0 700 900
4 =4 ID and 73=4 ID are the inside diameter of the window in inches.
a 3
Motor Protection 149

equipped as shown in Figure 9-6. This scheme conditions) in both the stator windings and rotor. The
combines excellent phase- and ground-fault sensitivity equivalent circuit during a locked-rotor condition is
with freedom from load current and starting current similar to a transformer-equivalent circuit with a
problems. resistively loaded secondary. The heat distribution
For high-resistance grounded systems, where very between the stator and rotor is contingent on the
high sensitivity is required, the CWP-1 directional relative stator resistance and 60-Hz rotor resistance.
ground relay should be considered. The voltage across Unlike an overload condition, in which heat can be
the transformer grounding resistor may be used as a absorbed over time by the conductors, core, and
voltage polarizing source (Fig. 9-7). The relay has a structural members, a locked-rotor condition produces
sensitivity of 7 mA at 69 V. significant heat in the conductors, which has little time
The maximum torque angle occurs when the current to be transferred to other sections of the motor.
leads the polarizing voltage by 458. It is interesting to Extreme heating takes place and can be tolerated by
note that the maximum sensitivity angle is leading the the motor for a very limited time. The time that a
reference voltage  3 V0 by 458. In high-resistance motor can remain at standstill after being energized
grounded systems, the predominant impedance for varies with the applied voltage and is an I2t limit. A
ground faults is the zero sequence network containing relay with an I2t characteristic that could be set for any
the grounding resistor as 3R and the zero sequence permissible locked-rotor times and locked-rotor cur-
distributed capacitance. The resulting fault current rents would naturally be the best choice for protecting
calculated is influenced by this RC circuit and thus the motor.
leads the applied voltage instead of lagging as would The heat generated within the motor can be
normally occur for a ground fault on a low-resistance approximated by
grounded system. The CWP-1 directional ground relay
is intended for use only on high-resistance grounded
I2H ¼ I21 þ KI22

4 LOCKED-ROTOR PROTECTION I1 ¼ per unit stator positive sequence current
K ¼ weighting factor to describe the increased
A rotating motor dissipates far more heat than a motor rotor resistance due to skin effect in the rotor
at standstill, since the cooling medium flows more bars at (2fs  fslip)
efficiently. During a failure to start or accelerate after I2 ¼ per unit stator negative sequence current
being energized, a motor is subject to extreme heating fs ¼ system frequency
(approximately 10 to 50 times more than for rated fslip ¼ slip frequency
Utilizing both the positive and negative sequence
currents in an equation relating to IH allows the motor
to be protected throughout the full range of stator
current with and without unbalance. The time-current
characteristic is as shown in Figure 9-8 for the MPR
relay. To insure adequate locked-rotor protection, the
curve position can be set slightly below the full-voltage
locked-rotor time. Depending on the availability of
RTDs, the cutoff can be set to protect the motor
during an overload condition. A typical cutoff setting
without RTDs would be 115 to 125% of full-load
If the time required for the motor to accelerate the
load is significantly less than the permissible locked-
motor time, the motor can be effectively protected
using conventional time-overcurrent relays (Fig. 9-10)
or microprocessor relays (Fig. 9-8). If, however, there
Figure 9-6 Flux balancing differential scheme. is little difference in the two time periods, or the
150 Chapter 9

Figure 9-7 Typical connections of the product type CWP-1 (32N) for high-resistance grounded systems.

starting time exceeds the locked-rotor time (Fig. 9-9), current drops below the CO pickup before area A
other considerations must be taken into account. equals area B.
For the case shown in Figure 9-9, it is tempting to When both voltage and current are available, an
try to fit an overcurrent relay characteristic between alternative solution to locked-rotor problems for
the two curves. It should be remembered, however, large motors is to use a distance relay and timer.
that the conventional overcurrent relay characteristic is A motor during start behaves like a three-phase
a plot of operating time against sustained current (Fig. balanced fault so a distance relay that responds to
9-10), whereas the starting characteristic is a trace of three-phase balanced faults should be used. The
current against time (Fig. 9-9). If ILR is applied to the impedance of the motor will remain fixed (largely
CO relay for time ta, the contacts are very nearly reactive at a low power factor) if the motor does not
closed. Current does not drop below the CO pickup accelerate. If the motor accelerates, both the impe-
value until time tb. Contact closure occurs at some dance and power factor will increase (Fig. 9-12). The
point tc even though the CO relay characteristic is impedance of a motor with a locked rotor is
always above the current trace. essentially independent of terminal voltage and, as
Over a narrow range, such as that between two and the motor accelerates, its impedance changes as
three times pickup, a CO relay can be assumed to indicated. This change of impedance with motor
operate if the integral of (I  1)n dt exceeds K, where I acceleration makes the distance relay particularly
is the multiple of pickup and n and K are constants well-suited to this application. It also affords backup
that depend on the relay type and time dial setting. If a protection for phase faults and some ground faults.
linear-linear plot of (I  1)n and t is used, a varying The timer can be obtained from one of two types of
current and time can be compared with the relay relays: a time-overvoltage relay (59) or time-over-
characteristic on an area basis. In Figure 9-11, for current relay (50), each supervised by the 52a breaker
example, the CO relay contact will not close if the contact.
Motor Protection 151

Figure 9-8 Typical MPR characteristics.

An alternative to the distance relay technique is the estimate the speed of the motor, and the rotor
PRO*STAR2 motor protection relay, which deter- thermal model then accounts for the declining rotor
mines the speed-dependent heating in the rotor during resistance during acceleration. Thus a high-inertia
a start. The relay uses an impedance measurement to motor where the allowable locked-rotor time is less
152 Chapter 9

Figure 9-9 Motor starting time exceeding permissible

locked rotor time.
Figure 9-11 Area comparison.

than the normal starting time can be properly

protected by this relay.
Some applications use a mechanical zero-speed
switch to supervise an overcurrent unit, preventing
operation of a timer once motor rotation is detected.
This scheme will not detect a failure to accelerate to
full speed nor pullout with continued rotation, as the
above schemes will.

Figure 9-12 KD-10 distance relay (21) used for locked rotor
Figure 9-10 CO characteristic compared to current trace. and backup protection for large motor.
Motor Protection 153


Heating curves are difficult to obtain and vary

considerably with motor size and design. Further, these
curves are an approximate average of an imprecise
thermal zone, in which varying degrees of damage or
shortened insulation life may occur. It is difficult, then,
for any relay design to approximate these variable
curves adequately over the range from light sustained
overloads to severe locked-rotor overload.
Thermal overload relays offer good protection for
light and medium (long-duration) overloads, but may
not for heavy overloads (Fig. 9-13a). The long-time
induction overcurrent relay offers good protection for
the heavy overloads, but overprotects for light and
medium overloads (Fig. 9-13b). A combination of the
two devices provides complete thermal protection (Fig.
The National Electric Code requires that an over-
load device be used in each phase of a motor ‘‘unless
protected by other approved means.’’ This requirement
is necessary because single phasing (opening one
supply lead) in the primary of a delta-wye transformer
that supplies a motor will produce three-phase motor
currents in a 2:1:1 relationship. If the two units of
current appeared in a phase with no overload device,
the motor would be unprotected. Thus, the NEC
requires three overload devices, or two overload
devices and another to detect unbalance such as a
CM or 46D relay.


There are two types of thermal relays. The DT-3 and

type 49T operate from a resistance temperature
detector (RTD) that monitors the temperature in the
machine windings, motor or load bearings, or load
case. They are typically applied only to large motors,
usually 1500 hp and above where RTDs are available.
The RTD is an excellent indicator of average winding
temperature. It is influenced by the effects of ambient
temperature, ventilation variations, and recent loading
The second type comprises replica relays such as the
BL-1, type 49, and IMPRS, which use an overcurrent Figure 9-13 Typical motor and relay time current char-
element with very slow reset characteristics to replicate acteristics.
the thermal condition of the motor and are applied
where RTDs are not available. use thermal replica algorithms. When RTDs are
Multifunction motor protection relays such as the available, the direct temperature measurement is also
PRO*STAR, MPR, and REM-543 include multiple used, and alarm and trip set points are established for
RTD inputs. When no RTDs are available, these relays each of the RTDs.
154 Chapter 9

No current-responsive relay can protect a motor The error associated with the resistance of the leads is
subjected to blocked ventilation. Relays using RTD removed and translations from resistance to tempera-
inputs for thermal protection overcome this short- ture can be performed by the microprocessor relay
coming by responding to temperature alone. using setting data stored in nonvolatile RAM or
EEPROM that relates to the physical properties of
the particular material used by the RTD.
The DT-3 is a bridge-type relay. The exploring coils
6.1 RTD-Input-Type Relays
form part of a Wheatstone bridge circuit, which is
balanced at a given temperature. As the motor
Several RTD types are available for use in temperature
temperature increases above the balance temperature,
operating torque is produced (Fig. 9-15). With the DT-
10 O copper 3 relay, only one resistance temperature detector (10,
100 O nickel 100, or 120 O) or exploring coil is required.
120 O nickel The DT-3 relay is a d’Arsonval-type dc contact-
100 O platinum making milliammeter that is connected across the
bridge. The bridge is energized by either 125 or
Microprocessor-based motor protection relays typi-
250 Vdc or supplied with 120 Vac through a transfor-
cally use three-wire input RTDs. The RTD has a well-
mer and full-wave bridge rectifier in the relay. The
defined ohmic characteristic vs. temperature. Accurate
relay scale is calibrated from either 50 to 190 8C (or 100
detection of the resistance of an RTD requires that the
to 1608). The right- or left-hand contacts close when
lead resistance be subtracted from the total resistance
measured by the relay. One scheme used by the MPR the temperature rises or falls to the preset value
between 50 and 190 8C (or 100 to 160 8C). The normal
circulates a precision current out the X terminal,
setting for class B machines is 120 8C.
through the RTD, and returns to terminal Z via the
return lead (Fig. 9-14). The following equation results:
VXZ ¼ RLEAD ICC þ RRTD ICC þ RLEAD ICC 6.2 Thermal Replica Relays
Replica-type relays (BL-1 and, additionally, the
RLEAD ¼ resistance of the leads to the RTD IMPRS, MPR, and PRO*STAR) are designed to
ICC ¼ constant current source replicate, within the relay operating unit, the heating
RRTD ¼ resistance of the RTD characteristics of the machine. Thus, when current
from the current transformer secondary passes through
It next circulates this same current through the Y
the relay, its time-overcurrent characteristic approxi-
terminal and return path to Z to obtain
By subtracting VYZ from VXZ, the relay obtains

Figure 9-15 Typical schematic of the type DT-3 relay (49)

for motor overload protection. (Its advantages are good
protection for overload, blocked ventilation, and high
Figure 9-14 MPR three-wire RTD input. ambient temperature operation.)
Motor Protection 155

mately parallels that of the machine capability curve at (27/47), or solid-state Types 27/27D and Types 47/47D
moderate overload. relays will accurately detect undervoltage and initiate a
Extreme variations in load, such as jogging, produce trip or alarm as required.
a difficult relaying problem. In general, electromecha-
nical thermal replica relays cool at a different rate from
the motor they protect. Variations in load may
produce a ratcheting effect on the relay and cause
premature tripping. Microprocessor motor protection
When starting in reverse can be a serious hazard, a
relays typically acknowledge previous loading history
reverse-phase relay should be applied. The relay, such
through the use of an RTD input, which establishes a
as types CP, CVQ, 47, or 60Q, monitors the bus
starting level.
voltage and is wired to supervise motor starting.
The thermal replica relay is recommended when
Another technique, used by the MPR multifunction
embedded temperature detectors are not available,
relay, is to trip instantaneously when reverse-phase
although RTD-input-type relays are recommended
currents are detected. The voltage relay method
when they are. Replica-type relays are typically
prevents motor energization, whereas the current
temperature-compensated and operate in a fixed time
method requires that the motor be energized and
at a given current, regardless of relay ambient
then tripped when reverse-phase sequence exists.
variations. Although this characteristic is desirable
for the stated conditions, it produces underprotection
for high motor ambient and overprotection for low

The CVQ (27/47) relay contains a negative sequence

voltage unit that operates as shown in Figure 9-16. A
negative sequence voltage network, as described in
Low voltage prevents motors from reaching rated
Chapter 3, energizes an induction-disk voltage unit V2.
speed on starting, or causes them to lose speed and
If a three-phase voltage applied to the relay contains
draw heavy overloads. An equation of the average
5% (adjustable to 10%) negative sequence content or
accelerating motor torque is directly related to the
more, the negative sequence unit (V2) operates. A back
voltage present:
TA ¼ Tm  TL
ER ¼ rated motor voltage
E ¼ voltage available at motor bus
TL ¼ load torque
Tm ¼ rated voltage motor torque
TA ¼ average accelerating torque
It can be seen that the voltage available to the motor
significantly affects the accelerating torque of the
Motors should be disconnected when severe low-
voltage conditions persist for more than a few seconds.
Use of ac contractors, which generally release at 50 to
70% of rated voltage, provides some low-voltage
protection. However, time-delayed undervoltage pro-
tection is preferred, since it delays contactor release on
momentary voltage dips. For switchgear applications, Figure 9-16 Simplified schematic diagram of the CVQ (27/
the electromechanical CV (27), CP (27/47), and CVQ 47) negative sequence voltage relay.
156 Chapter 9

contact of the negative sequence unit opens a CV-7

undervoltage unit coil circuit, and after a time delay,
the contacts of the undervoltage unit initiate tripping
or sound an alarm. This relay operates for

Reverse-phase rotation (100% negative sequence)

Unbalanced voltage (partial negative sequence)
Undervoltage (no negative sequence)

The CVQ relay is recommended for all important

buses supplying motor loads.
Although the CVQ relay can detect single phasing
of the supply to even a single, lightly loaded large
motor if its magnetizing impedance is low enough, it
does not respond to single phasing between the point
of application of the CVQ and the motor. Figure 9-17
displays two cases of an open phasing condition. The
first case is an open phase at A. The resulting sequence
network interconnections are shown.
In this first case, the negative sequence voltage relay
measures the voltage across the negative sequence
impedance of the motor or motors. In the second case,
the open phase occurs at B. Figure 9-17 shows the
sequence network interconnections. When the open is
Figure 9-18 External schematic of the CVQ relay used for
at location B, the relay now measures the negative tripping on negative sequence voltage only.
sequence voltage across the source.
Very low negative sequence voltage is produced on
the source side of the open phase, which makes it relay to detect. For practical purposes, the voltage
extremely difficult for the negative sequence voltage unbalance occurs only on the load side of an open
phase. In general, a phase-unbalance current relay is
preferred for detecting a feeder circuit open phase.
Figure 9-18 illustrates one type of CVQ relay that
responds only to negative sequence voltage (not under-
voltage). The six-cycle timer prevents operation for non-
simultaneous pole closure of the supply breaker, 52–1.
The solid-state relay types 47 and 47D are the
functional equivalents of the CVQ providing reverse-
phase unbalanced voltage and undervoltage protec-
tion. The type 60Q responds only to negative sequence
voltage, and will operate only for unbalanced condi-
tions. The 60Q includes an adjustable time delay.


Phase-unbalance protection is applied to a feeder

supplying a large motor or group of small motors
where there is a possibility of one of the feeder phases
opening as a result of a connector failure, fuse failure,
or similar cause. The electromechanical type CM
relay (device 46) contains two induction-disk units
Figure 9-17 Motor single phasing. (Fig. 9-19). One unit in the CM relay balances Ia
Motor Protection 157

against Ib, and the other balances Ib against Ic. When of unbalance are used to increase the estimated
the currents become sufficiently unbalanced, torque is temperature condition of the thermal replica (PRO*
produced in one or both of the units, closing their STAR, MPR, REM543).
contacts (which are connected in parallel in the trip
circuit). The solid-state type 46D determines the
negative sequence content of the three-phase currents
and includes a built-in timer. 11 NEGATIVE SEQUENCE CURRENT RELAYS
One relay can protect many motors subject to
collective single phasing. In addition, a phase-unbalance No standards have been established for the I22 t short-time
relay may protect up to five motors subject to individual capability for a motor, although I22 t ¼ 40 is regarded as a
single phasing, depending on how the motors are conservative value. (I2 is the per unit machine negative
operated and their relative sizes. For example, the relay sequence current, t the time in seconds.)
will not operate if a motor with a rating of one-fifth of The electromechanical negative sequence time-over-
the total feeder load is subject to single phasing while current relay COQ does not have the sensitivity
unloaded and while the remaining motors are fully necessary to properly protect a motor against the
loaded. The CM relay has 7-A continuous capability overheating caused by a prolonged load current
and operates when the unbalance exceeds approxi- unbalance. The CM and type 46D are the preferred
mately 10 to 15% between 2 and 7 A. With no current single-function relays. The phase-unbalance protection
present in one current circuit and with 1 A in the other, of the multifunction motor protection relays also
the relay operates (on the 1-A tap). provide the required sensitivity. The MPR can be set
Multifunction motor protection relays usually to pick up when I2 ¼ 5 to 30% of the full load tap setting,
include an element for phase unbalance protection. If and the IMPRS and PRO*STAR when I2 ¼ 10 to 50%.
the unbalance exceeds a set threshold for the set time Settings in the range 10–20% would be most typical.
delay period, tripping is initiated (PRO*STAR,
IMPRS, MPR, REM543). In addition, lower levels


A motor can experience excessive torque and over-

current in response to a jam condition that can be
caused by a binding action of the motor, bearings, or
driven load. To detect a jam condition, the relay has to
screen out other possibilities. The motor contribution
to a nearby fault, which can last for a few cycles, can
be screened out by setting the jam time delay greater
than the motor fault contribution time. High current is
not recognized as a jam condition unless the motor has
been through a start and is in a normal running state.
Multifunction relays usually include this element, as
they are capable of keeping track of the state of the
motor. The advantage of the jam function is much
faster tripping than would be provided by the locked
rotor protection.


The sudden reduction of shaft load is referred to as

load loss, which can be caused by a shaft breakage,
Figure 9-19 The type CM phase unbalance relay (46). loss of prime on a pump, or the shearing of a drive
158 Chapter 9

pin. To minimize any damage to the driven load, the

motor should immediately be taken offline. Detection
of load loss requires the recognition of the difference
between no load preceding the application of load and
no load following the application of load. The load
loss element is commonly supplied in multifunction
motor protection relays (MPR, PRO*STAR,
REM543) that have the ability to track the state of
the motor.

Figure 9-20 KLF used for motor loss of field detection.
Out-of-step protection is applied to synchronous
motors and synchronous condensers to detect pullout
resulting from excessive shaft load or too low supply
voltage. Other causes of pullout result from a fault
occurring on the supply system, whereby the fault type,
clearing time, and location are factors relating to the
stability of the motor. Underexcitation caused by
incorrect field breaker trip or a short or open in the
field circuit can also result in loss of synchronism of the
motor. For a discussion of the out-of-step protection
of large motors, refer to Chapter 8 on generator
Small synchronous motors with brush-type exciters
are often protected against out-of-step (or loss-of-
excitation) operation by ac voltage detection devices
connected in the field. No ac voltage is present when
the motor is operating synchronously.


Synchronous motors can be protected against loss of

excitation by a low-set undercurrent relay connected in
the field. This relay should have a time delay on
dropout to trip or alarm the operator. The KLF (or
KLF-1) relay (40) (described in Chap. 8) can also be
used to protect large motors against loss of field. The
under-voltage units of these relays should have their
contacts shorted. Loss of excitation of a synchronous
motor does not usually depress the voltage enough to
operate reliably an undervoltage unit.
Unlike undercurrent relays, the KLF (or KLF-1)
relay can detect both partial and complete loss of field,
and some out-of-step conditions as well (Fig. 9-20).
Both out-of-step and loss-of-excitation conditions
can be detected with a CW watt-type relay (55), Figure 9-21 CW watt relay used for out-of-step detection.
Motor Protection 159

Table 9-2 Typical Protection for Motors Below 1500 hp (Fig. 9-22)

Device number Quantity Description Typical setting Remarks

49/50 1 BL-1,2 unit with 2 ITs Set at full load for Good overload
motor with 1.15 protection.
service factor and
90% of full load for
1.00 service factor
motor. IT set 2 times
locked rotor.
51/50 1 CO-5, 1–12 A; with Current setting 1/2 Locked-rotor
IIT 10–40 A locked rotor. Time protection when
delay set to give starting time 20 to
CO-11, 4–12 A with operating time > 70 sec.
IIT 10–40 A starting time. IT set Locked-rotor
2 times locked rotor. protection when
starting time
20 sec.
50G 1 IT, 0.15–0.3 A single 0.15 A. For use with 50/5
unit BYZ.
47/27 1 CVQ, 5 to 10% V2 Low voltage 75 to Undervoltage, phase
sensitivity, 55- to 80%. V2 ¼ 5%. sequence, and
140-V range unbalanced voltage
51N/50N (alternative 1 CO-11, 0.5–2.5 A; with Pickup 0.5 A, time Provides ground
to 50G where BYZ IIT 10–40 A 0.1 sec at IT setting. protection. Time
cannot be applied) IT set 46IFL . unit overrides false
residual during

connected for 0 torque when the current lags the

voltage by an appropriate power factor angle, such
as 308. Used in this way, the CW is referred to as
a power-factor relay. The connection shown in
Figure 9-21 gives maximum contact closing torque
when the current lags its unity power-factor position
by 1208.


Table 9-2 and the associated Figures 9-22 and 9-23

show typical application combinations for motor
protection. Table 9-3 lists the combined protection
Figure 9-22 Motor protection below 1500 HP. functions that are available by using a microprocessor-
based motor protection relay.
160 Chapter 9

Figure 9-23 Motor protection 1500 HP and above.

Motor Protection 161

Table 9-3 Typical Protection for Motors of 1500 hp (Fig. 9-23)

Device number Quantity Description Typical setting Remarks

49 1 DT-3, 50–190 8C, Set for motor max. Overload protection:

specify ohms of safe operating blocked ventilation
RTD 120 vac. temperature. or high ambient.
51 1 CO-5, 1–12 A Current setting 1/2 Locked-rotor
locked rotor. protection when
starting time is 20 to
70 sec.
CO-11, 4–12 A Time delay setting to Locked-rotor
give operating protection when
time > starting time. starting time
20 sec.
50 1 SC, 2 unit, 20–80 A Set 2 times locked Fault protection.
50G 1 IT, 0.15–0.3 A single 0.15 A. For use with 50/5
unit BYZ.
46 1 CM, 1–3 A For IFL  3 A: Set 2A. Unbalanced current
For IFL < 3 A: Set 1A.
47/27 1 CVQ, 5 to 10% V2 Low voltage 75 to Undervoltage, phase
sensitivity, 55- to 80%, V2 ¼ 5%. sequence, and
140-V range unbalanced voltage
protection. Note:
CP volt relay can be
used in place of
CVQ if all three-
phase motors on bus
are protected by CM
87f 1 IT, 0.15–0.3 A, 3 unit 0.15 A. Provides phase and
ground protection.
Use three 50:5 BYZ
transformers. 50G
still required for
cable protection if
BYZ at motor.
87 (alternative to 50 3 CA, 10% None. Phase-fault protection.
and 87f—use where
minimum 3f fault
current available is
less than 5 times
motor starting
current and 87f
cannot be used)
51N/50N (alternative 1 CO-11, 0.5–2.5 A, with Pickup 0.5 A, time Provides ground
to 50G where BYZ IIT 10–40 A 0.1 sec at IIT setting. protection.
cannot be applied) IIT set 4 times full
Transformer and Reactor Protection
Revised By: J. J. McGOWAN


Differential relays are the principal form of fault When a transformer is first energized, a transient
protection for transformers rated at 10 MVA and magnetizing or exciting inrush current may flow. This
above. These relays, however, cannot be as sensitive as inrush current, which appears as an internal fault to
the differential relays used for generator protection. the differentially connected relays, may reach instan-
Transformer differential relays are subject to several taneous peaks of 8 to 30 times those for full load.
factors, not ordinarily present for generators or buses, The factors controlling the duration and magnitude
that can cause misoperation: of the magnetizing inrush are

Different voltage levels, including taps, that result in Size and location of the transformer bank
different primary currents in the connecting Size of the power system
circuits. Resistance in the power system from the source to
Possible mismatch of ratios among different current the transformer bank
transformers. For units with ratio-changing taps, Type of iron used in the transformer core and its
mismatch can also occur on the taps. Current saturation density
transformer performance is different, particularly Prior history, or residual flux level, of the bank
at high currents. How the bank is energized
308 phase-angle shift introduced by transformer
wye-delta connections.
Magnetizing inrush currents, which the differential
relay sees as internal faults. 2.1 Initial Inrush

Transformer protection is further complicated by a When the excitation of a transformer bank is removed,
variety of equipment requiring special attention: the magnetizing current goes to 0. The flux, following
multiple-winding transformer banks, zig-zag transfor- the hysteresis loop, then falls to some residual value fR
mers, phase-angle regulators (PAR), voltage regula- (Fig. 10-1). If the transformer were reenergized at the
tors, transformers in unit systems, and three-phase instant the voltage waveform corresponds to the
transformer banks with single-phase units. residual magnetic density within the core, there would
All the above factors can be accommodated by the be a smooth continuation of the previous operation
combination of relay and current transformer design, with no magnetic transient (Fig. 10-1). In practice,
along with proper application and connections. Mag- however, the instant when switching takes place cannot
netizing inrush, the most significant variable in be controlled and a magnetizing transient is practically
transformer protection, will be discussed first. unavoidable.

164 Chapter 10

Figure 10-1 Magnetizing current when transformers were

reenergized at that instant of the voltage wave corresponding
to the residual magnetic density within the core.

In Figure 10-2, it is assumed that the circuit is

reenergized at the instant the flux would normally be at Figure 10-2 Magnetizing current when transformers were
its negative maximum value ðfmax Þ. At this point, the reenergized at the instant when the flux would normally be at
its negative maximum value.
residual flux would have a positive value. Since
magnetic flux can neither be created nor destroyed
instantly, the flux wave, instead of starting at its
normal value ðfmax Þ and rising along the dotted line,
will start with the residual value (fR) and trace the
curve (ft).
Curve ft is a displaced sinusoid, regardless of the
magnetic circuit’s saturation characteristics. Theoreti-
cally, the value of fmax is þðjfR j þ 2jfmax jÞ. In
transformers designed for some normal, economical
saturation density fs , the crest of ft will produce
supersaturation in the magnetic circuit. The result will Figure 10-3 A typical magnetizing inrush current wave.
be a very large crest value in the magnetizing current
(Fig. 10-2).
The residual flux fR is the flux remaining in the core AIEE report, time constants for inrush vary from 10
after the voltage is removed from the transformer cycles for small units to as much as 1 min for large
bank. The flux will decrease along the hysteresis loop units.
to a value of fR , where i ¼ 0. Because the flux in each The resistance from the source to the bank
of the three phases is 1208 apart, one phase will have a determines the damping of the current wave. Banks
positive fR and the other two a negative fR , or vice near a generator will have a longer inrush because the
versa. As a result, the residual flux may either add to or resistance is very low. Likewise, large transformer units
subtract from the total flux, increasing or decreasing tend to have a long inrush as they represent a large L
the inrush current. relative to the system resistance. At remote substa-
A typical inrush current wave is shown in Figure tions, the inrush will not be nearly so severe, since the
10-3. For the first few cycles, the inrush current decays resistance in the connecting line will quickly damp the
rapidly. Then, however, the current subsides very current.
slowly, sometimes taking many seconds if the resis- In addition to the conditions that influence single-
tance is low. phase inrush, the wave shape of the inrush current into
The time constant of the circuit (L/R) is not, in fact, a delta winding is influenced by the number of cores
a constant: L varies as a result of transformer affected and the vector sum of the currents from the
saturation. During the first few cycles, saturation is bank windings. The net wave could, in fact, become
high and L is low. As the losses damp the circuit, the oscillatory (Fig. 10-4). The shape of a polyphase or
saturation drops and L increases. According to a 1951 single-phase inrush to a delta winding is affected by the
Transformer and Reactor Protection 165

The maximum line-current inrush occurs at +308

closing angles.
Because of the delta connection of transformer wind-
ing or current transformers, the maximum line-current
Figure 10-4 Typical magnetizing inrush current wave that
inrush value should be considered when applying
can exist in one of the phases to a delta connection or in the
current to the differential relay.
secondary of delta connected current transformers.

nature of the line current itself, which is the vector sum 2.2 Recovery Inrush
of two currents from the bank windings. If we assume
that only one core has saturated, the nature of the line An inrush can also occur after a fault external to the
current can result in either oscillatory waves or bank is cleared and the voltage returns to normal (Fig.
distortion of the single-phase shape. 10-5). Since the transformer is partially energized, the
When there is more than one delta winding on a recovery inrush is always less than the initial inrush.
transformer bank, the inrush will be influenced by the
coupling between the different voltage windings.
Depending on the core construction, three-phase
2.3 Sympathetic Inrush
transformer units may be subject to interphase
coupling that could also affect the inrush current.
When a bank is paralleled with a second energized
Similar wave shapes would be encountered when
bank, the energized bank can experience a sympathetic
energizing the wye winding of a wye-delta bank or an
inrush. The offset inrush current of the bank being
autotransformer. Here, the single-phase shape would
energized will find a parallel path in the energized
be distorted as a result of the interphase coupling
bank. The dc component may saturate the transformer
produced by the delta winding (or tertiary).
iron, creating an apparent inrush. The magnitude of
Maximum inrush will not, of course, occur on every
this inrush depends on the value of the transformer
energization. The probability of energizing at the worst
impedance relative to that of the rest of the system,
condition is relatively low. Energizing at maximum
which forms an additional parallel circuit. Again, the
voltage will not produce an inrush with no residual. In
sympathetic inrush will always be less than the initial
a three-phase bank, the inrush in each phase will vary
The maximum inrush for a transformer bank can be
calculated from the excitation curve if available, and
Table 10-1 shows a typical calculation of an inrush
current (used phase A voltage as 08 reference).
From these calculated values it can be seen that:

The lower the value of the saturation density flux

fS, the higher the inrush peak value.
The maximum phase-current inrush occurs at the 08 Figure 10-5 Recovery inrush after an external fault is
closing angle (i.e., 0 voltage). cleared.

Table 10-1 Typical Inrush Current Calculation

Peak value of inrush current wave (p.u.)

fs Closing angle Ia Ib Ic IaIb IbIc IcIa

1.40 08 5.60 3.73 3.73 8.33 3.73 8.33

1.40 308 5.10 1.87 5.10 5.96 5.10 9.20
1.15 08 6.53 4.67 4.67 10.20 4.67 10.20
1.15 308 6.03 2.80 6.03 7.83 6.03 11.06
166 Chapter 10

3.1 Differential Relays for Transformer

3.1.1 CA (87) Transformer Differential Relay
The CA transformer differential and the CA generator
differential are companion electromechanical relays.
Though they have largely been replaced by solid-state
and microprocessor relays, their operating principle is
still of interest. Since there are thousands of these
relays in service, they are included here.
Figure 10-7 shows the basic design of the transfor-
mer version of this relay. The generator relay has no
Figure 10-6 Sympathetic inrush when a bank is paralleled
with a second energized bank.
taps and is more sensitive than the transformer version.
The transformer relay is relatively insensitive to the
high percentage of harmonics contained in magnetiz-
ing inrush current. Because of this and its relatively
slow six-cycle operating time, it has been used
As shown in Figure 10-6, the total current at successfully for many decades in less critical applica-
breaker C is the sum of the initial inrush of bank A and tions where cost has a significant influence.
the sympathetic inrush of bank B. Since this waveform Operation occurs when the operating current (which
looks like an offset fault current, it could cause is the differential current) exceeds roughly 50% of the
misoperation if a common set of harmonic restraint minimum restraint. Putting it in another frame of
differential relays were used for both banks. reference, it operates when the operating current
Unit-type generator and transformer combinations exceeds approximately 20% of the summation
have no initial inrush problem because the unit is restraint. For external faults, note that the restraint
brought up to full voltage gradually. Recovery and ampere-turns are additive; and for an internal fault
sympathetic inrush may be a problem, but as indicated they are subtractive. For the idealized case of equal
above, these conditions are less severe than initial contributions from sources on each side of the
inrush. transformer to an internal fault, the restraint cancels



Since the differential relays see the inrush current as an

internal fault, some method of distinguishing between
fault and inrush current is necessary. Such methods

A differential relay with reduced sensitivity to the

inrush wave (such units have a higher pickup for
the offset wave, plus time delay to override the
high initial peaks), such as types of CA and CA-
26 transformer differential relays
A harmonic restraint or a supervisory unit used in
conjunction with the differential relay, such as
types of HU, HU-1, HU-4, TPU, and RADSB
transformer differential relays
Desensitization of the differential relay during bank
energization Figure 10-7 Type CA transformer differential relay.
Transformer and Reactor Protection 167

completely, and the currents add together in the for all such ‘‘through’’ phenomenon that provide this
operating coil. The transformer differential version of same ratio of currents, cancellation of operating torque
this relay allows currents with as much as a 2:1 ratio to occurs. Figure 10-9 shows a typical distribution of
be matched. Figure 10-8 describes the case with 10 A currents for a ‘‘through’’ condition with the relay set to
input from one set of current transformers and 5 A to balance currents with a 5 to 8 relationship.
the other set. Balance is accomplished in the operating
coil by the autotransformer action. For this case, and
3.1.2 CA-26 (87) Transformer Differential Relay
The CA-26 is a similar design to the CA-16 bus
differential relay, but with the ability to accommodate
all of the necessary input currents for protecting a
three-winding transformer. With no taps, the relay
requires auxiliary current transformers in two of the
input circuits to produce a match. One relay per phase,
of course, is required for a three-phase transformer.
Figure 10-10 describes the mechanical configuration
used with this relay and the principle by which
restraint is produced. It is sensitive and reasonably
fast, but its lack of harmonic restraint leads one to the
selection of more modern relays.

3.1.3 HU and HU-1 Transformer Differential

Figure 10-8 Distribution of currents in the type CA relay Relays
set on the 5–10 taps.
Since magnetizing inrush current has a high harmonic
content, particularly the second harmonic, this second
harmonic can be used to restrain and thus desensitize a
relay during energization. The method of harmonic
restraint is not without its problems. There must be
enough restraint to avoid relay operation on inrush,
without making the relay insensitive to internal faults
that may also have some harmonic content.
The HU (two restraining winding) and HU-1 (three
restraining winding) variable-percentage differential
relays have second-harmonic restraint supervision
Figure 10-9 Distribution of currents in the type CA relay that adequately solves these problems. The connec-
set on the 5–8 tap for example of Figure 10-14. tions for these relays are shown schematically in

Figure 10-10 Type CA-26 transformer differential relay.

168 Chapter 10

Figure 10-11 Schematic connections of the HU and HU-1 variable percentage differential relays with second harmonic
restraint supervision.

Figure 10-11. In the differential unit (DU), air-gap parallel, the relay restraint is proportional to the
transformers feed the restraint circuits, and a non-air- maximum restraining current in any restraint circuit.
gap transformer energizes the operating coil circuit. The percentage characteristic varies from around
Since the rectified restraint outputs are connected in 20% on light faults, where current transformer
Transformer and Reactor Protection 169

performance is good, to approximately 60% on heavy 3.1.5 Modified HU Relays

fault, where current transformer saturation may occur.
In the HU relay, the 15% second-harmonic restraint
This variable-percentage characteristic is obtained via
value is based on a minimum second-harmonic content
the saturating transformer in the operating coil circuit.
15% of fundamental, under an inrush condition of
Taps provide a 3:1 difference in the ratio of the main
fS ¼ 1:40 p:u: and a 08 closing angle. In modern
current transformer outputs. These taps are 2.9, 3.2,
transformers, however, saturation density is more
3.5, 3.8, 4.2, 4.6, 5.0, and 8.7.
often 1.20 to 1.30 p.u. and can even be as low as
The minimum pickup current is 30% of the tap
1.0 p.u. At these lower saturation densities, the
value for the 30% sensitivity relay and 35% of tap value
minimum second-harmonic content of the fundamen-
for the 35% sensitivity relay. The minimum pickup is
tal is significantly lower. Also, service conditions are
the current that will just close the differential unit
more severe when the closing angle is +30 , rather
contacts, with the operating coil and one restraint coil
than 08. (See Table 10-1.)
energized. The continuous rating of the relay is 10 to
As a result, the second-harmonic content percentage
22 A, depending on the relay tap used.
may be as low as 7% of the fundamental, and the
The harmonic restraint unit (HRU) has a second-
percentage of all harmonics may be as low as 7.5% of
harmonic blocking filter in the operating coil circuit
the fundamental. Under these conditions, the 15%
and a second-harmonic pass filter in the restraint
second-harmonic restraint relays may not function
coil circuit. Thus, the predominant second-harmonic
properly during the energization of power trans-
characteristic of an inrush current produces ample
formers with low saturation density values.
restraint with minimum operating energy. The
The modified HU relay is designed to solve this
circuit is designed to hold open its contacts when
problem. This relay is similar to the 15% second-
the second-harmonic component is higher than 15%
harmonic HU relay, except that a 33-O, 3-W resistor is
of the fundamental. This degree of restraint in the
connected across the HRU operating coil to calibrate
HRU is adequate to prevent relay operation on
the unit for a 7.5% second-harmonic restraint. The
practically all inrushes, even if the differential unit
characteristics of the modified HU are the same as for
should operate.
the HU relay, and the modification does not affect the
For internal faults, ample operating energy is pro-
characteristics of the IIT unit. The differential unit of
duced by the fundamental frequency and harmonics
the modified relay, however, is about one cycle slower
other than the second. The second harmonic is at a
than the unmodified unit.
minimum during a fault. Since the HRU will operate at
Ths modified HU relay has been used successfully in
the same pickup as the DU, the differential unit will
inrush current tests at a 138-kV generating station and
operate sensitively on internal faults, as shown in the
in several installations where the 15% HU relays
trip circuit of Figure 10-11. For external faults, the
experience inrush problems.
differential unit will restrain.
The relay operating time is one cycle at 20 times tap
3.1.6 Type RADSB Transformer Differential
value. The instantaneous-trip unit (IIT) is included to
ensure high-speed operation on heavy internal faults,
where current transformer saturation may delay HRU The type RADSB transformer differential relay is a
contact closing. The IIT pickup is 10 times the relay tap solid-state three-phase package. Its basic version
value. This setting will override the inrush peaks and provides two restraining circuits for two-winding
maximum false differential current on external faults. power transformer protection. It can be expanded up
to six restraints initially or in the future.
As shown in Figure 10-12, the relay utilizes the
3.1.4 HU-4 Transformer Differential Relay
second harmonic for inrush current restraint. The relay
The HU-4 relay is used to protect multiple-winding will restrain if the second harmonic content in any one
transformer banks or in the protective zone that phase is greater than 17% of its fundamental. This
includes the bus. The HU-4 relay is similar to the feature is very unique in three-phase package design. It
HU and HU-1 relays, but has four restraint windings. will solve the inrush current problem mentioned in
Also, the rectified outputs of the restraint transformers Section 3.1.5, ‘‘Modified HU Relays.’’
are connected in series, and the IIT unit is set at 15 The relay utilizes the fifth harmonic for the over-
times the tap value. The application of this relay is excitation restraint. The relay restrains if the fifth-
described in Chapter 11. harmonic content is greater than 38% of its funda-

Figure 10-12 Block diagram of RADSB transformer differential relay.

Chapter 10
Transformer and Reactor Protection 171

mental. However, refer to Section 5.5, ‘‘Overexcitation Fifth harmonic blocking is favored for those over-
Protection of a Generator-Transformer Unit,’’ in this voltage cases (long EHV line energization, hydroma-
chapter. It is necessary to apply the fifth harmonic with chine load rejection, etc.) where undesired high-speed
caution, or a V/Hz relay, either type MVH or tripping on elevated magnetizing current may occur. It
RATUB, should be considered for supervision. should be recognized that while transformers can
The relay does not provide built-in taps; therefore, it support short-time overvoltage, they are vulnerable
requires external auxiliary ct’s for current matching in to prolonged overvoltage heating. If differential trip-
all applications. Instantaneous high set trip function is ping is to be blocked when inordinate fifth harmonic is
also built in the relay. observed on overvoltage, the block must be released
For applications that required more than two (or tripping imposed by other means) prior to the
restraining circuits, the RTQTB-061 unit(s) can be occurrence of damage to the transformer.
added. Each RTQTB-061 unit contains six transfor- All-harmonic blocking is moderately less depend-
mers for two additional restraining circuits, as shown able because of the increased harmonics that are
in Figure 10-12. present in arcing faults.
Cross-blocking, the feature that blocks all differen-
tial tripping when the harmonic restraint setting and
3.1.7 TPU 2000R Transformer Protection System
the operating current are exceeded in any one or more
This unit takes full advantage of microprocessor phases, increases security against an unwarranted
technology. It allows the selection of a wide variety operation during inrush. This feature is partially
of characteristics such as second-harmonic restraint, inherent where ct’s are traditionally connected in delta
all-harmonic restraint, and fifth-harmonic restraint. for balancing the transformer phase shift and is useful
Inputs to it may be from wye- or delta-connected with the harmonic-restraint mode of second or second
current transformers, irrespective of the transformer and fifth. For an internal ground fault, possible
winding connections. This allows monitoring of the elevated voltage on an unfaulted phase should be
individual input phase currents rather than a combina- investigated to assure that excessive fifth harmonic in
tion of them, as a delta-connected set of ct’s would that phase cannot block tripping of the faulted phase
provide. The relay algorithm assures a proper match of differential unit. Where this can occur, cross-blocking
the input currents from both (or all three) sides of the must not be used.
transformer. It is compatible (different styles) with a
two- or three-winding transformer.
Any conceivable overcurrent unit curve shape can 3.2 General Guidelines for Transformer
be chosen as a setting for the overcurrent functions of Differential Relaying Application
this relay. Similarly, any of the communication
protocols that are in popular use can be provided as The following guidelines are designed to assist in
an inherent part of this device. Operating curve slope selecting and applying relays for transformer protec-
and minimum trip level are also obtained by setting. tion. When two or more relays appear to be equally
Inrush monitoring is possible, with blocking of suitable, engineering experience and economics will
tripping being selectable by second harmonic, fifth determine the final choice.
harmonic, or all harmonics (through the eleventh). 1. There is no clearcut answer to the question of
Historically, each concept has been used successfully. which relay or protective method to apply. As a
Though all of these methods of recognizing inrush general rule, however, the induction-disk differential
as a distinctive phenomenon for which trip blocking is relays (CA and CA-26) are used at substations remote
mandatory are useful, each has its own favorable and from large generating sources where inrush is not a
unfavorable nuances. Second-harmonic blocking is problem and the kVA size of the bank is relatively
minimally less secure for those cases involving over- small. The more complex and more expensive harmo-
voltage. The higher the voltage upon energization, the nic relays (HU, HU-1, HU-4, TPU and RADSB) are
lower the percentage of second harmonic. In general, used at generating stations and for large transformer
the relay is not capable of tripping at an overvoltage units located close to generating sources, where a
level below which the typical transformer can support severe inrush is highly likely.
on a prolonged basis. The relay expresses appropri- 2. A current transformer tap that will give
ately the need to trip. However, it would trip far too approximately 5 A at maximum load is recommended
soon. for use with multiratio current transformers. This
172 Chapter 10

arrangement provides good sensitivity without intro- 5. Relay taps or auxiliary current transformer
ducing thermal problems in the current transformer, ratios should be as close as possible to the current
leads, or relay itself. Sensitivity can be improved by ratios for a balanced maximum load condition. When
using a tap that gives more than 5 A; however, the there are more than two windings, all combinations
current transformer, leads, and relay capability must must be considered, two at a time, and based on the
be checked carefully to guard against thermal over- same kVA capacity.
load. 6. Ground only one point in the differential
3. In general, for all except the TPU, the current scheme; never do multiple-point grounding.
transformers on the wye side of a wye-delta bank must 7. After the current transformer ratios and relay
be connected in delta, and the current transformers on taps have been selected, the continuous rating of relay
the delta side connected in wye. This arrangement (1) windings should be checked for compatibility with the
compensates for the 308 phase-angle shift introduced transformer load. If the relay current exceeds its
by the wye-delta bank and (2) blocks the zero sequence continuous rating, a higher-current transformer ratio
current from the differential circuit on external ground or relay tap may be required. The relay’s required
faults. As shown in Figure 10-13, zero sequence current continuous rating may be determined from the
will flow in the differential circuit for external ground maximum kVA capacity of the transformer bank. If
faults on the wye side of a grounded wye-delta bank; if the transformer is allowed to exceed its maximum kVA
the current transformers were connected in wye, the capacity for a short time, the expected 2-h maximum
relays would misoperate. With the current transfor- load should be used. The relay will reach final
mers connected in delta, the zero sequence current temperature within 2 h.
circulates inside the current transformers, preventing 8. The percentage of current mismatch should
relay misoperation. always be checked to ensure that the relay taps selected
4. Relays should be connected to receive ‘‘in’’ and have an adequate safety margin. When necessary,
‘‘out’’ currents that are in phase for a balanced load current mismatch values can be reduced by changing
condition unless the relay itself is designed or set to current transformer taps or adding auxiliary current
accommodate the difference. When there are more transformers. Percentage mismatch M can be deter-
than two windings, all combinations must be consid- mined from Eq. (10-1):
ered, two at a time. I 
 L  TL 
M¼ 6100% ð10-1Þ

IL, IH ¼ relay input currents, at the same kVA base,
for low- and high-voltage sides, respectively
TL, TH ¼ relay tap settings for low- and high-
voltage sides, respectively
S ¼ smaller of the two terms, (IL/IH) or (TL/
When there are more than two windings, all combina-
tions should be calculated, two at a time. When taps are
changed under load, the relays should be set on the basis
of the middle or neutral tap position. The total mis-
match, including the automatic tap change, should not
exceed the recommended values shown in Table 10-2.
For example, for a transformer bank with a +10%
on-load tap changer device, the calculated mismatch
value should not be greater than +5% for a 30% HU
relay application. However, if the transformer bank
does not have an on-load tap changer, then the
Figure 10-13 Reason for delta-connected ct’s on wye calculated mismatch value can be tolerated up to the
windings. ‘‘limit of (M þ LTC)’’ value.
Transformer and Reactor Protection 173

Table 10-2 Recommended Mismatch (M) Limitation Table 10-3 Multiplier (m) for Eq. (10-3)

Limit of 3fF fGF

Sensitivity (M þ LTC)
Relay (%) (%) Wye-connected ct m¼1 m¼2
Delta-connected ct m¼3 m¼2
CA 50 35
HU, HU-1, HU-4, TPU 30 15
HU, HU-1, HU-4, TPU 35 20
CA-26, RADSB — 10

The following examples show the importance of the

9. To ensure correct operation of the relaying current transformer connections, current ratios, relay
scheme, the current transformer performance should ratings, and current transformer performance in
be checked. A less accurate, but still acceptable, applying the differential relay scheme for transformer
method is to use the ANSI relaying accuracy classifica- protection.
tion. For units with class C accuracy, performance will
be adequate if
4.1 Checks for Two-Winding Banks
Np VCL  ðIext  100ÞRS
> ZT ð10-2Þ A worksheet for connecting differential relays around
a two-winding bank is shown in Figure 10-14. This
where does not apply to the TPU relay.

Np ¼ proportion of total current transformer turns 4.1.1 Phasing Check

in use, for example, if 1000/5 tap is used for a
2000/5 MR current transformer, then It is very important to note that the transformer bank,
Np ¼ 0.5 as shown in Figure 10-14, is connected so that the high
VCL ¼ current transformer accuracy, class C vol- side lags the low side by 308, which is not an ANSI-
tage; for example, 200 for a class C200 standard-connected bank. (In a standard connection,
current transformer the high side leads the low side by 308.) In this
Iext ¼ maximum external fault current in secondary example, the nonstandard connection is used for
rms A (let Iext ¼ 100 if maximum external illustration only. Practically, the actual connection of
fault current is less than 100 A) the bank should be confirmed with the information
Rs ¼ current transformer secondary winding resis- from its nameplate before proceeding to the next step
tance in ohms of the phasing check.
ZT ¼ total current transformer secondary circuit Procedures for phasing check can be simplified as
burden impedance in ohms, determined by below (refer to Figures 10-14 and 10-15).
equation: Step 1 Assume that Ia , Ib , and Ic on the wye side
flow through the bank to an external three-phase
ZT ¼ 1:13ðm6RL Þ þ relay burden þ ZA ð10-3Þ
fault or maximum load.
where Step 2 See Figure 10-14; use the statement on page
12 to define the current in the windings: ‘‘the
RL ¼ one-way lead resistance current flowing out at the polarity-marked
1.13 ¼ multiplier used to accommodate temperature terminal on the secondary side is substantially
rise of the conductors during faults in phase with the current flowing in at the
ZA ¼ burden impedance of any devices (other than polarity-marked terminal on the primary side.’’
the relay) connected or reflected to the Three-phase transformers do not carry polar-
current transformer secondary circuit ity marks as do their single-phase counterparts.
m ¼ multiplier, depending on the current trans- However, from the nameplate data for the
former connection and type of fault to be transformer, it can clearly be seen which of the
considered, as shown in Table 10-3 low-voltage windings is drawn in parallel with
174 Chapter 10

Figure 10-14 Worksheet for connecting differential relays around a two-winding transformer bank.

one of the high-voltage windings. This is a ‘‘A-phase’’ on the low-voltage side will be
symbolic identification that these two windings connected to the X1 terminal, but surprisingly
are on the same core-leg. If these windings this is seldom the case. This simply adds to the
(actually lines) are vertical on the nameplate complexity of the relay engineer’s task, sorting
diagram, it may be assumed that the polarity out all of these factors. In fact, the transformer
markings could be placed at the upper end of connection dictates the relationship between the
each of these windings. Similar identification can high- and low-voltage systems, and the phasors
be made of the other windings. The terminals of a can be named anything with which the user feels
three-phase transformer are identified as H1, H2, comfortable.
H3, and, if there is a neutral, H0 (note this says Step 3 Trace these currents through the delta to
terminals, not windings). The low-voltage term- the delta-side phase wires, then through the wye-
inals are called X1, X2, and X3 (and possibly connected current transformers to the relays.
X0). If there is a third set of windings and the Step 4 Repeat the above relay currents to the
terminals are brought out of the case, they will be other-side restraint windings.
identified as Y1, Y2, and Y3 (and again Y0 if Step 5 Trace the currents on the wye-winding
there is a neutral and it is brought out of the phase wires; then determine the secondary
case). currents (phase and direction) on each wye-side
The nameplate will show whether the voltage current transformer.
drop from H1 to H3 is in-phase with X1 to X0 Step 6 Match up the information from steps 4 and
(high leads low by 308), or H1 to H2 is in-phase 5, enabling the wye-side current transformers to
X1 to X0 (high lags low by 308) or some other be properly connected in delta for correct
combination. It is assumed that the system phasing under all conditions. The completed
‘‘A-phase’’ on the high-voltage side will be check for the example of Figure 10-14 is shown
connected to the H1 terminal and that the in Figure 10-15.
Transformer and Reactor Protection 175

Figure 10-15 Complete phasing check for the example of Figure 10-14.

4.1.2 Ratio Check Step 8 Connect relay terminal 9 to the 69-kV side
and terminal 7 to the 11.5-kV side. Always set the
For the example in Figure 10-14, the ratio check
time dial at position number 1.
should be executed as shown in Table 10-4. The steps
Step 9 Check the continuous rating of the relay
that follow depend on the type relay being applied: CA
coils. As shown in calculations, the continuous
(steps 7 to 10); CA-26 and RADSB (steps 11 to 13); or
currents flowing in the restraint coils are less than
HU (steps 14 to 17):
10 A, and any through currents flowing in the
operating coil are less than 5 A. Therefore, the
relay windings will not be subject to a thermal
4.1.3 For Type CA Relays (Steps 7 to 10) problem.
Step 10 Calculate mismatch; use Eq. (10-1). This
For the application of CA relays, the tap settings, mismatch is well within the 35% mismatch limit
continuous coil ratings, and mismatch must also be of Table 10-2.
checked as described in steps 6 to 9. The taps in CA
relays are 5-5, 5-5.5, 5-6.6, 5-7.3, 5-8, 5-9, and 5-10.
Tap ratios are 1.00, 1.10, 1.32, 1.46, 1.60, 1.80, and
2.00, respectively. 4.1.4 For Type CA-26 and RADSB Relays (Steps
11 to 13)
Step 7 Select relay taps with a ratio as close as
possible to the relay current ratio in step 5. In this To calculate mismatch for CA-26 relays, perform steps
case, tap 5-8, with a ratio of 1.60, is the closest. 11 to 13 as follows:
176 Chapter 10

Table 10-4 Example of Ratio Check for Two-Winding Transformer

Step LV (wye) HV (delta)

1. For the example shown in Figure 10-14, assume that the maximum load ffiffi
p30;000 ¼ 1506 A ffiffi
p30;000 ¼ 251 A
is 30,000 kVA. Then the rating of the bank IFL is 3611:5 3669:0
2. For increased sensitivity, select current transformer ratios as close to the
IFL value as possible. Practically, a calculated value of
(IFL/0.8) can be used as the reference for determining the current 1506 251
transformer ratios for this example. 0:8 ¼ 1882:5 0:8 ¼ 313:7
2000 300
Then use n¼ 5 n¼ 5
3. Calculate current transformer secondary currents IS ¼ (IFL/n) ¼ 400 ¼ 60
1506 251
¼ 400 ¼ 60
¼ 3.77 Apffiffiffi ¼ 4.18 A
4. Calculate relay current IRL ¼ 3:77ð 3Þ IRH ¼ 4.18 A
¼ 6.52 A
IRL 6:52
5. Calculate relay current ratio IRH ¼ 4:18 ¼ 1:560

Step 11 Calculate mismatch; use Eq. (10-1). If a current-balancing autotransformer is used,

refer to Figure 10-16 and apply Eq. (10-1):
M ¼ 4:18 6100%
S 4:17
1:560  1 M ¼ 4:18 6100% ¼ 0:24%
¼ 6100% ¼ 56% 1
Step 12 Since the percent mismatch is higher than Both values are well within the 10% mismatch
the recommended value (Table 10-2), an auxili- limit of Table 10-2.
ary current transformer or a current-balancing
autotransformer is required to decrease the
current to the relay. (Note: Use auxiliary current
transformer to decrease the current to the relay.) 4.1.5 For Type HU Relays (Steps 14 to 17)
The turns ratio of the balancing current trans- To apply HU relays, the tap settings, continuous coil
former is (4.18/6.52) 6 100%, or 64.1%. Either a ratings, and mismatch must be checked, as described in
two-winding auxiliary current transformer with a steps 13 to 16. The taps in HU relays are 2.9, 3.2, 3.5,
turns ratio of 3/2 (for example, ABB style no. 3.8, 4.2, 4.6, 5.0, and 8.7. The tap ratios are given in
7881A026G06), or a current-balancing auto- Table 10-5.
transformer (for example, ABB type A auxiliary
current transformers) would be satisfactory. Step 14 Select relay taps that have a ratio as close
The connections for these two transformers are as possible to the relay current ratio in step 5. In
shown in Figure 10-16. The continuous rating of this case, tap 5.0/3.2, with a ratio of 1.563, is the
the auxiliary current transformers should be closest.
checked to guard against possible thermal Step 15 Use tap 3.2 on the 69-kV side and tap 5.0
problems. on the 11.5-kV side. In most HU applications,
Step 13 After selecting the auxiliary current trans- auxiliary current transformers are not required
former ratio, the mismatch should be rechecked. for current balancing if the current ratio is
If a 3/2 ratio auxiliary current transformer is used, between 1 and 3.
refer to Figure 10-16 and apply Eq. (10-1): Step 16 Check the continuous thermal rating of
the relay coils. Since the continuous rating of the
1 relay is 12 A for the 3.2 tap, there should be no
M ¼ 4:18 6100% ¼ 3:80%
1 thermal problem with the relay coils.
Transformer and Reactor Protection 177

Step 17 Calculate mismatch:

M ¼ 4:18 3:2
1:560  1:563
¼ ¼ 100% ¼ 0:2%
This value is well within the 15% mismatch limit
of Table 10-2. Also, the 30% sensitivity of the
HU relay would be satisfactory for this applica-

4.1.6 For Type TPU Relays

The TPU is applied similarly to the HU relays with
added benefits. Its wider tap range and finer tap
increments yield an even smaller mismatch than that
achieved with the HU relay. Also, the ct’s can be
connected in wye on both sides of the transformer,
with compensation for transformer phase shift accom-
plished internally for phase current metering and
overcurrent protection on the wye side of the
transformer. For the example in Table 10-4, the tap
calculations and selections
pffiffiffi would be the same as that
for the HU relay. The 3 factor for IRL would still
apply because of the TPU internal compensation and
not because of delta-connected ct’s.

4.1.7 Current Transformer Performance Check

Assume that the three-phase external fault currents are
higher than the single-phase fault currents, with values
of 15,000 A on the 11.5-kV side and 2500 A on the 69-
kV side. In this case, the current transformer burden
limit can be calculated as follows:
Figure 10-16 Current balancing transformer connections
for Figure 10-15 when type CA-26 or RADSB relay is used.
Low High
voltage voltage

Maximum external fault 15,000 2500

current (primary
Table 10-5 HU Relay Tap Ratios amperes), Ip
Current transformer turns 400 60
2.9 3.2 3.5 3.8 4.2 4.6 5.0 8.7 ratio, n
2.9 1.000 1.103 1.207 1.310 1.448 1.586 1.724 3.000 Secondary amperes
3.2 1.000 1.094 1.188 1.313 1.438 1.563 2.719 37.5 41.7
n 2000 300
3.5 1.000 1.086 1.200 1.314 1.429 2.486
3.8 1.000 1.105 1.211 1.316 2.289
3000 ¼ 0:67 600 ¼ 0:50
4.2 1.000 1.095 1.190 2.071
4.6 1.000 1.087 1.890
5.0 1.000 1.740
From Eq. (10-2) where (Iext  100)RS ¼ 0 since Iext is
8.7 1.000
less than 100 A secondary (Figure 10-14 shows the
178 Chapter 10

C400 current transformers are used in this example), in the other winding. Any other pair is then checked in
the same manner. This process ensures that all ratios
0:676400 0:56400
¼ 2:67 O ¼ 2:0 O are correct for any distribution of fault or load current.
100 100 A worksheet showing the connection of differential
Current transformer performance will be satisfactory if relays around a typical three-winding bank is given in
the total burden impedance values, as calculated from Figure 10-17. The completed phasing checks are shown
Eq. (10-3), are less than the above values. in Figure 10-18.
If the ratios are not correct for the relay, auxiliary
current-balancing autotransformers or current transfor-
4.2 Checks for Multiwinding Banks mers are required. In general, one or two sets are required
for the three-winding bank, depending on the unbalanced
The same types of phasing, ratio, continuous rating, condition and relay type (CA-26, HU-1, or RADSB).
and current transformer performance checks are used For the three-winding bank shown in Figure 10-18,
for multiwinding transformer as for two-winding the ratio check is performed per Table 10-6, steps 1 to 5.
transformers. To determine the correct direction and
phase of the restraint currents, one side of the
4.2.1 For Type CA-26 and RADSB Relays (Steps
transformer is considered the primary and the other
6 to 8)
windings the secondaries. For ratio checks, any two
windings must be checked based on a same kVA value, For CA-26 relay application, mismatch is checked
as if the bank were a two-winding unit with no current according to steps 6 to 8 below.

Figure 10-17 Worksheet for connecting differential relays around a three-winding transformer bank.
Transformer and Reactor Protection 179

Step 6 Calculate mismatch; use Eq. (10-1):

MHM ¼ 6:012 6100% ¼ 26%
MML ¼ 4:380 6100% ¼ 37:3%
MHL ¼ 4:380 6100% ¼ 73%
Step 7 To decrease the currents to the relay, add
current-balancing autotransformers at the 66-
and 26-kV sides. The turn ratios of these
transformers are
6100% ¼ 57:8%
Use 73 turns ratio:
6100% ¼ 72:8%
Use 58 turns ratio:

Then the currents to the relay are

IRH ¼ 7:577658% ¼ 4:395 A

IRM ¼ 6:012673% ¼ 4:389 A
IRL ¼ 4:380 A
Figure 10-18 Complete phasing check for the example of
Step 8 Recalculate the mismatch:
Figure 10-17. The dotted lines show the connections for a
phasing check between the 66 and 11 KV windings. Assum- 4:395
ing the 26 KV circuit does not exist. The dashed lines show MHM ¼ 4:389 6100% ¼ 0:1%
the connections for a phasing check between the previous 1
connections made for the 66 KV winding, assuming that the 4389
11 KV circuits do not exist. With this method, phasing is MML ¼ 4:380 6100% ¼ 0:2%
correct for any distribution of currents through the three
MHL ¼ 4:380 6100% ¼ 4:4%

Table 10-6 Example on Ratio Check for Three-Winding Transformer

Step 66 kV (Y) 26 kV (Y) 11 kV (D)

1. If 25,000 kVA flows through pffiffi

¼ 219 pffiffi
¼ 556 pffiffi
¼ 1314
3666 3626 3611
the bank, the currents in
each winding are IFL
2. If we assume current n ¼ 250/5 ¼ 800/5 ¼ 1500/5
transformer turn ratios of ¼ 50 ¼ 160 ¼ 300
3. Then the current ¼ 219/50 ¼ 556/160 ¼ 1314/300
transformer secondary ¼ 4.380 ¼ 3.475 ¼ 4.380
currents are IFL/n pffiffiffi pffiffiffi
4. Relay currents are IRH ¼ 364:380 ¼ 7:577 IRM ¼ 363:475 ¼ 6:012 IRL ¼ 4:380
IRH 7:577 IRM 6:012 IRH 7:577
5. Relay current ratios are IRH ¼ 6:012 ¼ 1:260 IRL ¼ 4:380 ¼ 1:373 IRL ¼ 4:380 ¼ 1:730
180 Chapter 10

4.2.2 For Type HU-1 Relays (Steps 9 to 12) employed in electromechanical or solid-state relays.
Differential comparisons are made between the I/tap
For HU-1 relay application, tap settings, mismatch,
values of all currents entering and leaving the relay.
and current transformer performance are calculated as
The taps are chosen to be proportional to the (actual
or adjusted) input currents to the relay for a ‘‘through’’
Step 9 To select the relay taps, use Table 10-5 and fault (or load). Contrary to the traditional method of
start from the highest current ratio, IRH/ selecting the current transformer connections to deliver
IRL ¼ 1.730. The nearest tap ratio is 5/ currents to the relay with appropriate phase relation-
2.9 ¼ 1.724. Select TH ¼ 5 and TL ¼ 2.9. Next, ships, with this relay flexibility exists in those connec-
select the tap ratio for the second higher-current tions. Factors are applied in setting the relay that
ratio, IRM/IRL ¼ 1.373, by using the lower tap compensate for the connections of the transformer and
from the first tap ratio (TL ¼ 2.9) as a reference. the ct’s on the high and low (and third winding where
This ratio would be 3.8/2.9 ¼ 1.310. In other present) sides of the protected transformer. The
words, TM ¼ 3.8. selection is based on the angle by which the winding-
Step 10 Calculate the mismatch: 1, typically high voltage, input current leads the
comparable winding-2, typically low voltage (and
 5 winding-3, typically tertiary voltage, where applicable)
MHM ¼ 6:012 3:8 6100% ¼ 4:4%
1:26 outgoing current.
 3:8 The differential slope (IOP/IR) of the relay is linear
MML ¼ 4:380 2:9 6100% ¼ 4:8%
1:31 (15 to 60%) with a minimum operating current. Both
 5 slope and minimum trip level are selectable. Typically,
MHL ¼ 4:380 2:9 6100% ¼ 0:3% a slope of 20 or 30% is selected, depending on whether
or not the transformer is equipped with a load-tap
Step 11 Check current transformer performance in changer. A minimum operating current of 20 to 40% of
the same way as for the two-winding bank above. tap value may be chosen, with 30% being a traditional
Step 12 Even though the current ratios are within and successful setting.
3.0 (step 5), auxiliary current transformers may The relay system provides load current metering
be required for current balancing. For example, if (including peak demand), menu-driven programming
IRH ¼ 3.75, IRM ¼ 8.109, and IRL ¼ 6.222, the of settings, three groups of setting tables, program-
current ratios are 2.162, 1.303, and 1.659. Even mable input and output contacts (with and without
though these values are within 3, auxiliary time delay), oscillographic data storage, fault records,
current transformers are necessary in this appli- and communications ports.

4.2.3 For Type TPU Relays

Again, the TPU would be applied similarly to the HU PROTECTION
relay. In the three-winding TPU, p allffiffiffi ct’s must be
connected in wye, resulting in a 3 factor being 5.1 Differential Scheme with Harmonic
applied to IRH and IRM due to internal compensation Restraint Relay Supervision
as they were applied with the HU due to the ct’s being
connected in delta. By setting this relay with second- and fifth-harmonic
blocking, it will refrain from tripping on the disparity
between incoming and outgoing currents that result
4.3 Modern Microprocessor Relay from inrush or exciting current due to overvoltage.
Any combination of connections of a transformer can
The modern microprocessor relay, such as the TPU be accommodated. Contrary to conventional practice,
relay, is a protection system, rather than a simple the current transformers should be connected in wye
transformer differential overcurrent relay. Settings are on each side, irrespective of the phase shift between the
required for phase and ground (time and instanta- currents on one side of the transformer and those on
neous) as well as negative sequence overcurrent the other. Figure 10-19 describes a typical external
functions. For the differential function, the tap ranges connection of the current transformers for a TPU
are wider and the steps are smaller than those 2000R relay.
Transformer and Reactor Protection 181

Figure 10-19 Typical external connection for TPU relay.

The appropriate phase shift is taken into considera- mended. This preserves the proportionality between
tion internally in the relay. The currents that are the input and output currents and is consistent with the
delivered to the relay are modified in the setting other combinations. Similar settings to those applied
process to cause them to appear to have the same value for two-winding transformers are used for three-
they would have had if conventional ct connections winding transformers.
had been used. This modifier is called the compensa-
tion factor. Figure 10-20 shows the majority of
connections that are in use, the appropriate phase- 5.2 Ground Source on Delta Side
angle settings, and the compensating factors that
should be used. Consider, for example, that a delta As shown in Figure 10-21a, the differential relay will
connection of the ct’s were in use in the conventional operate falsely on external ground faults if the
setup. The current delivered to the pffiffiffirelay would be differential zone covers a grounding bank and a
multiplied, by the connection, to be 3 times the phase conventional wye-connected current transformer set
current. Using wye currents only instead of the is used. This misoperation can be eliminated by
conventional delta currents, the pffifficurrents
ffi are multi- inserting a zero sequence current trap in the circuit
plied, in the setting process, by 3 (the compensating (Fig. 10-21b). The zero sequence current trap consists
factor) to obtain equivalency. This internal compensa- of wye-delta-connected auxiliary current transformers,
tion is only applied to the phase differential algorithm. which can have any ratio. These auxiliary current
With the wye-wye connection, where the conventional transformers provide a low-impedance path for the
practice would dictate the ct’spbe
ffiffiffi connected
pffiffiffi in delta, zero sequence current component and high-impedance
the compensating factors of 3 and 3 are recom- path for the positive and negative sequence current
182 Chapter 10

Figure 10-20 TPU settings dictated by transformer connections.

Figure 10-21 Ground source on delta side.

Transformer and Reactor Protection 183

components of the fault current. The scheme can be flashovers if the power system is ungrounded, as
grounded at the differential relay or trap, but only one illustrated in Figure 10-22b. The relay scheme also
ground point may be used. will not detect an internal bushing flashover and
external ground fault on another phase.
Protection for these internal faults is obtained by
5.3 Three-Phase Banks of Single-Phase Units placing current transformers on both bushings of a
single-phase transformer winding that forms part of
Figure 10-22 shows one phase of the transformer the three-phase delta connection (Fig. 10-22c). All
differential connection for transformer bushing current current transformers can be wye-connected.
transformers used in a three-phase bank of single- An unbalance of 2/1 results from connecting the two
phase units. In such cases, conventional current current transformers in the delta winding in parallel
transformer connection cannot be used in the three with one ct in the wye winding. When selecting current
circuits to the delta, such as those shown in Figure 10- transformer ratios and/or relay taps, this unbalance
15. This differential relaying scheme will not detect must be taken into account.
internal bushing flashovers if the power system is
grounded as illustrated in Figure 10-22a. This differ-
ential relaying scheme will not detect internal bushing

5.4 Differential Protection of a Generator-

Transformer Unit

In the unit-type system shown in Figure 10-23, the

transformer differential relay is often connected to
include the generator as well as the transformer. This
arrangement provides additional and overlapping
protection for the rotating machine. Separate current
transformers on the generator neutral are recom-
mended to keep the burden low.
Because the station service transformer bank is
much smaller than the generator unit, the transformer
differential relay may not protect against secondary or
internal faults in the station service transformer unless
such faults occur near the high-voltage end of the
primary winding. The main units require a high current
transformer ratio to limit secondary currents under
continuous operation and high faults. Since the station
service unit is small, it will have a high impedance and
light fault currents—frequently below the transformer
differential relay sensitivity. In this case, a separate
differential relay around the station service bank can
be operated with current transformer ratios appro-
priate to the size of the bank. Overload relays, without
a separate differential, may be used to protect this
station service bank.
If no appreciable current is fed back through the
small station service unit for faults external to the large
main unit, faults on the low side of the station service
unit may be well below the main differential relay
sensitivity. If so, no current transformer connection to
Figure 10-22 Protection problems and solution for internal the main differential relay circuit is required. Other-
faults on delta side of a three-phase bank consisting of single- wise, a connection is required to prevent tripping on
phase units. station service bus faults.
184 Chapter 10

Figure 10-23 Differential protection of the unit-type generator-transformer system with separate differential protection for the
station service unit.

5.5 Overexcitation Protection of a Generator- C37.106—1987 ‘‘IEEE Guide for Abnormal-Fre-

Transformer Unit quency Protection for Power Generating Plants’’).
In a generator-transformer unit system, the trans-
Overexcitation of a transformer may result in thermal former may be subjected to an overvoltage or over-
damage to cores due to excessively high flux in the excitation condition on load rejection or as external
magnetic circuits. Excess flux saturates the core steel faults are cleared by the high-side breaker. During
and flows into the adjacent structure, causing high periods of high overexcitation, conventional transfor-
eddy current losses in the core and adjacent conducting mer differential relays and relaying schemes may
materials. operate. Some users, however, consider this ‘‘mis-
Since flux is directly proportional to voltage and operation’’ an advantage, since it protects against
inversely proportional to frequency, the unit of transformer damage from overvoltage.
measure for excitation is defined as per unit voltage Figure 10-25 shows a scheme for preventing unde-
divided by per unit frequency (volts/hertz). Over- sired tripping of the differential relay and of protecting
excitation exists whenever the per unit volts/hertz a transformer-generator combination against over-
exceeds the design limits of the equipment; for voltage. The differential relay must be equipped with
example, a transformer designed for a voltage limit some form of restraint that will recognize excessive
of 1.2 per unit at rated frequency will experience volts/hertz. Either fifth-harmonic or ‘‘all-harmonic’’
overexcitation whenever the per unit volts/hertz restraint is available for this function.
exceeds 1.2. Should the voltage exceed 120% at rated Figure 10-24 provides some help in establishing
frequency, or the frequency go below 83.3% at rated permissible limits for transformers, but the individual
voltage, overexcitation will exist. Severe overexcitation manufacturer should be contacted for assurances.
can cause rapid damage and equipment failure. In Figure 10-25, device 87 is chosen to provide
Figure 10-24 shows the curves of transformer over- protection to both the transformer and generator
excitation limitations for various manufacturers (and provide backup to the generator differential
(curves are from Figure 11 of ANSI/IEEE Standard relay, not shown). As voltage increases, due to, for
Transformer and Reactor Protection 185

time-delayed to coordinate with the capability of the

transformer and generator.
Care must be exercised in choosing both the level of
fifth harmonic at which blocking takes place as well
as the level of volts/hertz and time at which tripping
takes place. Information such as that provided in
Figure 10-26 should be obtained from the individual
manufacturer to identify what may be expected for a
particular transformer or generator.

5.6 Sudden-Pressure Relay (SPR)

With the application of a gas-pressure relay, many

transformers can be protected by a simple differential
relay set insensitively in the inrush current. The
sudden-pressure relay (SPR), which operates on a
rate of rise of gas in the transformer, can be applied to
any transformer with a sealed air or gas chamber
above the oil level. The relay is fastened to the tank or
manhole cover, above the oil level. It will not operate
on static pressure or pressure changes resulting from
normal operation of the transformer.
The SPR relay is recommended for all units of
5000 kVA or more. It is extremely sensitive to light
faults as it will operate on pressure changes as low as
0.33 lb/in.2/sec. In one case, this represented a fault of
50 A. The SPR relay is far more sensitive to light
internal faults than the differential relay. The differ-
ential relay, however, is still required for faults in the
bushing and other areas outside the tank.
The SPR relay operating time varies from one-
half cycle to 37 cycles, depending on the size of the
In the past, large-magnitude through-fault condi-
tions on power transformers have caused rate-of-
change-of-pressure relays to occasionally operate
falsely. There has been reluctance on the part of
Figure 10-24 Transformer overexcitation limitations for some users to connect these rate-of-change-of-pres-
various manufacturers transformer under no-load condi- sure relays to trip, and they have therefore used
tions. them for alarming only. Schemes have been devised
to restrict tripping of the rate-of-change-of-pressure
device only to levels of current below which the
transformer differential relay cannot operate. One
example, load rejection, the tendency of the differ- means of doing this is with a high-speed current-
ential relay to operate is blocked by, say, the blocking type RAICB relay (or a simple overcurrent
detection of increased fifth harmonic. As the voltage relay) to supervise the SPR trip circuit. By wiring
increases farther, to the point where damage may the SPR output to a TPU 2000R input, recording,
occur to the protected apparatus, device 24 (volts/ supervision, and time delay of an SPR operation
hertz relay) takes over to provide tripping that is can be accomplished.
186 Chapter 10

Figure 10-25 Overexcitation protection for generator and transformer.

5.7 Overcurrent and Backup Protection tertiary windings may carry very heavy currents during
ground faults. In such cases, tertiary overcurrent
To allow transformer overloading when necessary, the protection must be provided.
pickup value of phase overcurrent relays must be set A through fault external to a transformer results in
above this overload current. An inverse-time charac- an overload that can cause transformer failure if the
teristic relay usually provides the best coordination. fault is not cleared promptly. It is widely recognized
Settings of 200 to 300% of the transformer’s self-cooled that damage to transformers from through faults is the
rating are common, although higher values are some- result of thermal and mechanical effects. The thermal
times used. Fast operation is not possible, since the effect has been well understood for years. The
transformer relays must coordinate with all other mechanical effect has recently gained increased recog-
relays they overreach. nition as a major concern of transformer failure. This
Overcurrent relays cannot be used for primary results from the cumulative nature of some of the
protection without the risk of internal faults causing mechanical effects, particularly insulation compres-
extensive damage to the transformer. Fast operation sion, insulation wear, and friction-induced displace-
on heavy internal faults is obtained by using instanta- ment. The damage that occurs as a result of these
neous trip units in the overcurrent relays. These units cumulative effects is a function of not only the
may be set at 125% of the maximum through fault, magnitude and duration of through faults, but also
which is usually a low-side three-phase fault. The the total number of such faults.
setting should be above the inrush current. Often, The transformer can be isolated from the fault
instantaneous trip units cannot be used because the before damage occurs by using fuses or overcurrent
fault currents are too small. relays. The former ANSI C37.91, ‘‘Guide for Protec-
An overcurrent relay set to protect the main tive Relay Applications to Power Transformers,’’ was
windings of an autotransformer or three-winding based on the former U.S. Standard C57 for Power
transformer offers almost no protection to the tertiary Transformer, in which mechanical effect was not
windings, which have a much smaller kVA. Also, these considered. The latest published Standard C57.109–
Transformer and Reactor Protection 187

Figure 10-26 Example of harmonic components on transformer overexcitation.

1985, ‘‘IEEE Guide for Protective Relay Applications

Table 10-7 Transformer Category (ANSI/IEEE Standard to Power Transformers,’’ considers both the thermal
C57.109-1985 Curves) and mechanical effects.
For purposes of coordination of overcurrent pro-
Minimum nameplate (kVA)
Reference tective devices, ANSI/IEEE Standard C57.109–1985
Single- Three- protective presents different curves for different-size transformers
Category phase phase curve as listed in Table 10-7. For applications including the
category I transformer, only the thermal effect from
I 5–500 15–500 Fig. 10-27 the through-fault current is considered in Figure 10-27;
II 501–1667 501–5000 Fig. 10-28
on the contrary, for applications including the category
III 1668–10,000 5001–30,000 Fig. 10-29
IV transformer, the thermal and mechanical effects
IV above 10,000 above 30,000 Fig. 10-30
from the through-fault current should be considered as
188 Chapter 10

Figure 10-27 Through-fault protection curve for category 1 transformers 5 to 500 kVA single-phase 15 to 500 kVA three-phase.

shown in Figure 10-30. For applications including the some other fashion, the through-fault protection curve
category II or III transformer, Figures 10-28 and 10- should reflect primarily thermal damage considera-
29, whether or not the mechanical effect from the tions, since the cumulative mechanical damage effect
through-fault current should be considered depends on of through faults will not be a problem.
the frequency of the external fault. For applications in which external faults occur
For applications in which external faults occur frequently, for example, transformers with secondary-
infrequently, for example, transformers with second- side overhead lines, the through-fault protection curve
ary-side conductors enclosed in conduit or isolated in should reflect the fact that the transformer will be
Transformer and Reactor Protection 189

Figure 10-28 Category II transformers 501 to 1667 kVA single-phase 501 to 5000 kVA three-phase.

subjected to the thermal and cumulative mechanical The following example describes the procedures for
damage effects of through faults. constructing the dog-leg portion of the thermal/
Figure 10-31 shows the infrequent-frequent fault mechanical limit curves: a 230/25-kV, 30/50-MVA
incidence zones for the determination of curve selec- transformer with an impedance of 10% on a 30-MVA
tion. base and with secondary-side overhead lines.
190 Chapter 10

Figure 10-29 Category III transformers 1668 to 10,000 kVA single-phase 5001 to 30,000 kVA three-phase.

Step 1 Select the category from the minimum Step 2 Plot the infrequent through-fault curve of
name-plate rating of the principal winding. For category III (Fig. 10-29), as shown in Figure 10-
this example, it is a category III transformer, and 32, portion A.
the Figure 10-29 curve should be used for the Step 3 Determine the dog-leg portion of the curve
coordination. as follows:
Transformer and Reactor Protection 191

Figure 10-30 Category IV transformers above 10,000 kVA single-phase above 30,000 kVA three-phase.

1. Calculate the maximum per unit through- This is point 1 in Figure 10-32.
fault current: 2. Calculate the constant K ¼ I2 t at t ¼ 2 sec:
1 1 2
I¼ ¼ 106base current at 2 sec K¼ 62 ¼ 200
0:10 0:10
192 Chapter 10

Figure 10-31 Infrequent-frequent fault incidence zones for category II and category III transformers.

3. Calculate the time at 50% (note: use 50% for when the fault current flows toward the protected
category III and IV, 70% for category II) of transformer. They are set to reach into, but not
the maximum per unit through-fault current: beyond, the transformer.
K 200
t¼ 2
¼ ¼ 8 sec
I ½0:5ð10Þ2
This is point 2 in Figure 10-32. 5.9 Overcurrent Relay with HRU Supplement
4. Connect points 1 and 2 and draw a vertical
line from point 2 to the infrequent curve to Three single-phase HRU units with instantaneous
complete the dog-leg portion of the curve as trip elements can be used to supplement the time-
shown in Figure 10-32. delay overcurrent relays when inrush is a problem
and no current transformers are available on the
secondary side of the protected bank. As illustrated
5.8 Distance Relaying for Backup Protection in Figure 10-33, this arrangement provides high-speed
tripping when the transformer is energized on faults.
Directional distance relaying can be used for transfor- The scheme is not recommended, however, unless
mer backup protection when the setting or coordina- transformer loads are supervised by individual local
tion of the overcurrent relays is a problem. The breakers, or load pickup does not occur during
directional distance relays are connected to operate transformer energization.
Transformer and Reactor Protection 193



The protection of industrial and commercial power

transformer banks is somewhat different from the
protective schemes used by utilities. The differences in
protective schemes are a function of several major
factors, including system configuration, method of
grounding, speed, coordination, operation, and cost.
Some of the more commonly used industrial and
commercial protective schemes are shown in Figures
10-34 to 10-39.
Figure 10-34 illustrates how a primary breaker can
be used for transformer protection. The basic protec-
tion is provided by the 87T transformer differential
relays. Either type TPU, CA, or HU relays can be
used, depending on the severity of inrush and
operating speed requirements. Device 50/51, an
inverse-time CO relay with IIT unit, provides trans-

Figure 10-32 Multiple of transformer full load (per unit).

Figure 10-34 Transformer protection with primary

Figure 10-33 Overcurrent relay with single-phase HRU breaker. The MSOC relay may be used for the overcurrent
supplement for speed improvement. functions.
194 Chapter 10

Figure 10-35 Paralleled transformer protection with primary breaker.

Figure 10-36 Connections and operation at the CWC (87TG) ground differential relay where the connected system is
Transformer and Reactor Protection 195

Figure 10-37 Connections and operation of the CWC (87TG) ground differential relay where the connected system is
ungrounded or the external ground source is not always available.

former primary winding backup protection for phase provides feeder ground backup, whereas device 63,
faults; either device 50G (type ITH with a zero such as a type SPR relay, offers highly sensitive
sequence current transformer) or 50N/51N can be protection for light faults.
used as transformer primary winding backup for The current transformer ratings in this scheme
ground faults. Transformer overload, low-voltage should be compatible with the transformer short-time
bus, and feeder backup protection are provided by overload capability: approximately 200% of transfor-
device 51 on the transformer secondary side. Since the mer self-cooled rating for pffiffiffi wye-connected current
low-voltage side is medium-resistance-grounded, a transformers and 350% ð 36200%Þ for delta-con-
ground relay (51G) should be used to trip breaker nected current transformers. The neutral current
52-1 for low-side ground faults and for resistor thermal transformer rating should be 50% of the maximum
protection. Device 151G, which trips breaker 52–11, resistor current rating.
196 Chapter 10

sensitivity can be obtained by using a lower-ratio

neutral current transformer. It is desirable to keep the
currents in the two relay coils within a 2:4 ratio so that
an auxiliary current transformer will be required if a
large ground resistor is used.
Figure 10-37 shows the connections when the
external system is not grounded or the external ground
source is not always available. Iq must be positive for
the external fault. This can be done by using

n ¼ 1:2 or higher

for the auxiliary ct ratio as shown in Figure 10-37a.

Current transformer ratios, as well as any effect of
saturation of the line current transformer, must be
In addition to devices 51G and 151G shown in
Figure 10-35, a 251G relay is used to trip breaker 52T
on ground faults. The trip sequence of these three
ground relays is as follows: (1) 251G trips 52T, (2)
151G trips 52-11, (3) 51G trips 52-1. The 87TG trips
52-1 and 52-11. Device 67 (type CR relays) provides
reverse overcurrent protection.
Figure 10-38 Transformer protection with primary fuses. If the transformer is too small to warrant the basic
schemes described above, the scheme shown in Figures
10-38 and 10-39 is recommended. Here, fuses provide
When a normally closed secondary bus tie breaker the primary fault protection. Solid grounding will
is used for paralleled transformer protection (Fig. 10- assure sufficient primary phase fault current to operate
35), there are several differences, with the primary the fuses for most secondary ground faults. The
breaker scheme shown in Figure 10-34. First, a type opening of a single primary fuse will result in single
CWC relay (87TG) provides selective and sensitive phasing of the transformer secondary system. This
protection for ground faults within the secondary may be difficult to detect, particularly at light loads,
circuit of the differential zone. The CWC relay, an and appropriate precautionary measures should be
induction-disk relay, has two windings that operate taken.
on the product of the two currents. The operating If the primary source is grounded and there is a
torque is proportional to the product times the cosine power source on the secondary side, a ground fault on
of the angle between the two currents. Maximum the incoming line will be interrupted by the source
torque occurs when the currents are in phase; the breaker; the transformer primary or secondary
connections and operation of the scheme are shown in breaker, however, will not be relayed open because
Figures 10-36 and 10-37. of the delta primary transformer connection. The
For an external fault, the two currents in the relay failure of these breakers to open can result in hazards
coils are essentially 1808 out of phase. In this case, the to personnel, possible damaging transient overvoltages
CWC relay has no operating torque. For an internal produced by an arcing-type fault, and problems with
fault, the currents in the relay are essentially in phase, automatic reclosing of the source breaker. Several
producing operating torque. The relay sensitivity is schemes can be used to ensure opening of the source
0.25 VA. Make sure that the external ground source is and transformer breakers, including pilot protection of
always available when applying Figure 10-36; other- the incoming line, transfer-trip, or potential ground
wise, the CWC relay will not operate on internal detection relaying schemes on the transformer pri-
faults. mary. Automatic reclosing is a special problem,
In these applications, the ratios of the current requiring that the secondary breaker be opened before
transformers do not have to be identical. Increased the primary source breaker is reclosed.
Transformer and Reactor Protection 197

Figure 10-39 Paralleled transformer protection with primary fuses.

When a normally open bus tie breaker is used, as in 8 PROTECTION OF PHASE-ANGLE

Figures 10-35 and 10-39, devices 67 and 67N are not REGULATORS AND VOLTAGE
required. REGULATORS

A phase-angle regulating transformer, which inserts or

impresses a regulated voltage on a line in quadrature
7 REMOTE TRIPPING OF TRANSFORMER with its line-to-ground voltage, is used to control
BANK power flow in the system. A voltage-regulating
transformer compensates for drops in IR by inserting
Transformer banks are often applied as a part of the or impressing a regulated voltage on a line in phase
line section, with no high-side breakers. The protection with its line-to-ground voltage. These two transformers
problems associated with this combination are consist basically of a series unit and an exciting unit,
described in Chapters 12, ‘‘Line and Circuit Protec- located on at least two separate cores and in separate
tion.’’ tanks. Depending on the design and size of the bank,
198 Chapter 10

the exciting unit may be wye- or delta-connected, and In application, the current transformer ratio and
the series unit may be constructed as one unit or split relay tap’s selection for 87E would be similar to the
into two identical units. By mixing the control elements conventional differential scheme for three-winding
in the exciting unit, as shown in Figure 10-40, a single transformer protection. The use of an equal current
bank can sometimes provide both the phase-angle transformer ratio for the source, load, and primary
regulation and voltage control functions. The protec- of the exciting unit is recommended. This allows
tive schemes for the phase-angle regulator and voltage the use of equal tap settings in the differential
regulator are as varied as the ways in which the bank is relay. For most applications, the current transfor-
constructed. Figure 10-41 shows a typical scheme that mer ratio is determined by the full-load current
could be used for the system depicted in Figure 10-40. through the series winding. However, for larger-
Device 87E (HU-1 relays, one per phase; TPU angle shifts, the current transformer ratio is
relays, one per bank; or KAB relays, one per phase) determined by the current of the exciting-unit
provides overall regulator protection. As shown in primary winding.
Figure 10-42, the currents IS, IL, and Ie are applied To set the 87E relays at their minimum tap, current
vectorially to the relay. Since all these currents flow in transformer ratios should be identical. For HU-1 relay
the primary circuit of the series windings, saturation of application, a setting of 2.9 for all restraint elements
the series windings from external-fault overvoltages will provide the best sensitivity. The series-unit
will not affect the 87E relays. primary and exciting-unit primary are electrically

Figure 10-40 Typical 400 MVA 115 KV phase-angle regulator, +268 with voltage control.
Transformer and Reactor Protection 199

Figure 10-41 Typical scheme for protecting the phase-angle regulator of Figure 10-40.

connected (instead of magnetically), and there is no ð 3Þ to operate the relay and two out of the three
phase shift. Consequently, the current transformers for relays in the scheme will pick up on an internal fault
87E relays can be either wye- or delta-connected. Wye providing redundant backup.
connection will permit faulted-phase identification, Device 87S (HU relays, one per phase or TPU)
whereas delta connection will produce more current provides differential protection for the series unit. As
shown in Figure 10-44, the current transformers for
current Ie’ should be located at the neutral end of the
windings to provide some backup protection for the
exciting unit.
The series winding has very low impedance and is
designed for rated voltage equivalent to the quadrature
voltage at maximum phase-angle shift. For the
example shown in Figures 10-43 and 10-44, it is
approximately [2 sin (268/2)] or 45% of the line-to-
neutral voltage. As a result, the transformer windings
are subject to over-voltage or overexcitation conditions
on external fault, which may produce saturation of the
Figure 10-42 Overall differential protection for the phase- series windings and cause false operation of the 87S
angle regulator of Figure 10-40. relays.
Chapter 10

Figure 10-43 Phase-angle regulator (87S).

Transformer and Reactor Protection 201

Use the maximum load-current value to determine

the source-/load-side ct ratios:

315; 000
¼ 528 A

That is, use the 1200:5 ratio for the series-unit primary-
side ct’s:

Series-unit primary current ¼ 528 A

Series-unit secondary ¼ 1274 A
current ¼ 528 (2 6 280/232)
Exciting-unit secondary current ¼ 2204 A
Figure 10-44 Differential protection for series winding of ¼ 1.73 6 1274
the phase-angle regulator of Figure 10-40.
Therefore, the ct ratio at this location should not be
lower than 2204/0.8 ¼ 2755 to 5, i.e., use the 3000/5 ct.
87S relay taps selection would be as follows:
Whether or not an external fault will cause such
overvoltage on the winding and false relay operation
Series-unit primary-side ct ¼ 1200/5 delta
depends on several factors, including the characteristic
Exciting-unit secondary-side ct ¼ 3000/5 wye
of the series winding (saturation curve, impedance, and
Current from series-unit primary ¼ 3.804
tap position), location and type of fault, and power-
to relay ¼ 1.73(528)(5/1200)
system condition. A typical analysis shows that faults
Current from exciting-unit ¼ 3.673
on one side of the bank produce false operation of the
secondary to relay ¼ 2204
relay, whereas faults on the other side of the bank do
not. Other cases indicate no overvoltage problems at
Current ratio ¼ 2 6 3.804/3.673 ¼ 2.071
all. If overvoltage is a problem, therefore, the 87S
relays should be supervised by volts/hertz or excessive
Note the following:
fifth harmonic.
Note in Figure 10-43 that for this phase-shifting-
1. A factor of 2 is included in the current’s ratio
regulator protective scheme that the currents are
calculation for this particular example. This is
summed in the relay at the right. As shown in this
different from the approach in a conventional
drawing, the current in R2 (restraint 2) is equal to the
differential scheme.
sum of the currents in restraints R1 and R3. Differences
2. The following illustration shows a simpler way
in the tap settings are required only if the RCL does not
for finding the current’s ratio:
equal RCE (nt/ny).
For illustration purposes, a typical example of
current transformer ratio and relay tap selection for ne ¼ 232 Rce ¼ 600
the 87S device is given below. ns ¼ 280 RcL ¼ 240
It should be noted that the ampere turns in the ne 6Rce 2326600
series winding, primary and secondary, are always Current ratio ¼ ¼ ¼ 2:071
ns 6RcL 2806240
balanced for any ‘‘through’’ condition.
Selected relay taps’ ratio ¼ 8:7=4:2 ¼ 2:070
An example illustrates the 87S relay taps selection
for a phase shifter as shown in Figure 10-43. It has the Calculated mismatch ¼ < 0:01%
following information:
Use 8.7 taps for the series-unit ct’s and 4.2 tap
345 kV, plus/minus 608 189/252/315 MVA for the exciting-unit ct.
Series winding primary 2 6 280 turns In the event that ct ratios are to be
Series winding secondary 232 turns determined, the following expression may be
Exciting winding 334 turns helpful. Make M% approach 0; then the ct
202 Chapter 10

ratios can be determined: particular phase shift to provide a path for circulating
ne 6Rce
zero sequence current, it will provide no zero sequence
ns 6RcL  TTLe current for external ground fault. Therefore, the 51N1
M% ¼ ! 0:0
S and 51N2 devices do not have a coordination problem
where on external faults. However; coordination should be
considered if there is a zero sequence current path for
M% ¼ percent mismatch external ground faults in this area.
TL ¼ relay line/restraint tap setting The sudden-pressure relays are recommended for
Te ¼ relay exiting-unit/restraint tap setting these units, especially when there is high potential for
ne ¼ secondary series-winding number of turns an arcing fault in the tap-changing equipment.
ns ¼ half-primary series-winding number of
RcL ¼ line/source ct ratio
Rce ¼ secondary of exciting-unit ct ratio 9 ZIG-ZAG TRANSFORMER PROTECTION
S ¼ smallest of the two ratios
Since the connected system will be ungrounded in some
Devices 51N1 (short time) in the neutral circuit of the applications, a zig-zag grounding transformer can be
exciting-unit secondary provides sensitive ground fault protected against ground faults by the scheme shown
protection for single-phase-to-ground faults on the in Figure 10-48. The overcurrent relays for the delta-
secondary side of the exciting unit. The zero sequence connected current transformers provide phase-fault
current distribution for a ground fault in this area is protection (Fig. 10-48). The time-overcurrent relay
shown in Figure 10-45. (device 51N) in the neutral provides backup ground
Device 51N2 provides backup for devices 51N1 and protection. The ground relay must be set to coordinate
87E during single-phase-to-ground faults on the with ground relays in the connected system. Rate-of-
exciting-unit primary. The current flow in neutral pressure-rise relays, such as the sudden-pressure relays,
depends on the autotransformer action of the faulted are recommended for light internal faults.
winding (Fig. 10-46). The grounding banks are seldom switched by
Refer to Figure 10-47. Since there is no delta- themselves. When they are switched, however, they are
connected winding in the exciting unit for this subjected to magnetizing inrush—just as for other types

Figure 10-45 Sensitive ground protection zero sequence currents for a secondary ground fault in the exciting unit of the phase-
angle regulator of Figure 10-40.
Transformer and Reactor Protection 203

Figure 10-46 Ground backup protection and zero sequence currents for a primary ground fault in the exciting unit of the
phase-angle regulator of Figure 10-40.

of transformers. The harmonic restraint relay (single- tion increases as the square of the voltage and is a
phase-type HRU), as shown in Figure 10-48, can be used function of line length and the conductor configura-
to prevent inadvertent tripping during energization. tion. In many cases, it is necessary to absorb these
Some power transformer banks consist of zig-zag- VAR and provide voltage control at both terminals
connected windings for phase correction or system during normal operation. High overvoltage on
grounding. As shown in Figures 10-49 and 10-50, the sudden loss of load must be limited as well. System
phase-angle shift between the primary and secondary switching and operation may require a different
sides of the banks depends on their connections. In amount of VAR absorption and even, at times,
these examples, windings on one side are delta- some VAR generation.
connected. As shown in Figure 10-51, however, a Shunt reactance for VAR control is obtained by
wye connection could also be used, introducing a
Fixed shunt reactors
phase-angle shift of either plus or minus 308. The
Switched shunt reactors or capacitors
grounded wye is not a zero sequence current source for
Synchronous condensers
ground faults on the wye side, even if both windings
Static VAR compensators
are grounded. It is, however, a good zero sequence
current source for ground faults at the zig-zag side. Fixed shunt reactors are generally used for EHV and
Both the phase-angle shift and zero sequence current long HV lines and for HV cables. Switched shunt
source should be considered when applying the reactors or capacitors and synchronous condensers
differential scheme for these transformers. are applied in the underlying system and near load
Shunt reactors vary greatly in size, type, construc-
tion, and application. Their capacities range from 3 to
125 MVA, at voltage levels from 4.6 to 765 kV. They
10.1 Shunt Reactor Applications can be single- or three-phase, oil- or dry-type, with
either air or gapped-iron cores. The connections may
Both EHV transmission lines and long HV transmis- be directly (1) to the transmission circuit, (2) to the
sion lines and cables require shunt reactance to tertiary winding of a transformer bank that is part of
compensate for their large line-charging capacitance. the line, or (3) to the low-voltage bus associated with
This capacitance produces VAR generation that the the line transformer bank. (This third application is
system generally cannot absorb. This VAR genera- not common.)
204 Chapter 10

Figure 10-47 The exciting unit provides no zero sequence current path for external ground fault. Devices 51N1 and 51N2 do
not have coordination problem on external faults.

Line reactors, which are connected directly or Line-connected reactors are generally included
through a disconnect switch, are a part of the within the line protection zone and are often well
transmission circuit. Circuit breakers are seldom protected by the line relays adjacent to the units.
used. The neutrals of the reactors are solidly grounded Separate reactor relays are recommended, however,
or grounded through a neutral reactor. Reactor faults since the remote terminal may have difficulty detecting
require that all line terminals be open. a reactor fault. These relays can be applied with
When connected to the tertiary of a transformer current transformers sized to the reactor MVA and
bank, circuit breakers are generally used, either in the should include some way of transfer-tripping the
supply or on the neutral. Opening the neutral breaker remote line terminals—especially on long lines or
does not isolate a reactor fault. Tertiary applications when the remote terminal is a relatively weak source.
are operated either ungrounded or grounded through With separate reactor relays, the line relays provide
impedance. additional backup.
Line operation without a reactor can result in very Tertiary-connected reactors can be included in the
high overvoltage when load is lost, such as when one transformer bank differential zone. Separate reactor-
end is opened. This factor encourages the use of direct- protection relays are recommended. When practical,
connected reactors to avoid the accidental loss of the transformer protection zone overlap should be
service should load be lost. used as backup. Line-side reactor breakers allow the
Transformer and Reactor Protection 205

Figure 10-48 Protection of a zig-zag grounding transformer and the zero sequence currents for an external ground fault.

protection to be separated, so that the transformer Tripping is recommended, although such protection is
bank need not be tripped for reactor faults. In such sometimes used for alarm purposes only. An alarm
cases, the possibility of high voltage during operation operation should be monitored carefully since there are
without the reactors should be examined. cases where a fault left no tangible evidence after the
The protective techniques commonly used for first pressure relay operation but later developed into a
reactor primary and backup protection are severe fault. Even on the severe fault, the pressure relay
was distrusted because of the initial assumed-false
Rate-of-rise-of-pressure (applicable to oil units with
a sealed gas chamber above the oil level)
Rate-of-rise-of-pressure protection can be used as
Overcurrent (three-phase and/or ground)
separate primary protection only if line or transformer
Differential (three-phase or ground only)
differential protection is available for faults outside the
Other protective relaying techniques, such as distance, reactor tank and for backup protection. Rate-of-rise-
negative sequence, and current balance, have been used of-pressure protection is, of course, not applicable to
to a limited extent. dry-type units.

10.2 Rate-of-Rise-of-Pressure Protection 10.3 Overcurrent Protection

Rate-of-rise-of-pressure protection provides the most Overcurrent phase and ground protection for reactors
sensitive protection available for light internal faults. are shown in Figure 10-52. To avoid operation on
206 Chapter 10

Figure 10-49 Interconnected delta zig-zag transformers with voltages in phase on the two sides.

transients, the phase-type CO time-overcurrent units phase reactor units. With single-phase units, the
(51) are set at 1.5 times the rated shunt reactor separate differential relays aid in identifying the fault.
current: the IIT instantaneous units (50) are set at five The relays detect both winding and bushing faults. Since
times the rated current. The ground relay unit (51N) the relays will see magnetizing inrush as a ‘‘through’’
can be set at 0.5 to 1.0 A and the relay (50N) at five condition, generator-type relays can be used. Either the
times the 51N setting. Both ground units should be set SA-1 or generator CA-type relays may be applied; both
above the zero sequence current (3I0) contribution of provide sensitive internal fault protection (0.14 A for the
the reactor for faults outside the reactor protection SA-1 and 0.18 A for the CA). The ground relay (50N/
zone. This setting will avoid operation on line- 51N) provides backup protection.
deenergized oscillations. If the reactor is connected A single CA-16 or HU-4 relay can be used for a
to an ungrounded system, 50N and 51N should be ground differential. The four restraints are connected
omitted. This scheme requires only one set of current to the three lines and one neutral current transformer,
transformers. as shown in Figure 10-54. The minimum pickup of the
CA-16 is 0.15 A and of the HU-4 is 0.87 A.
The scheme shown in Figure 10-55 provides an
10.4 Differential Protection excellent combination of phase instantaneous and time
overcurrent with ground differential. For single-phase
Separate-phase differential relays (87), as shown in reactors, phase faults that do not involve ground
Figure 10-53, are applicable for either three- or single- cannot occur—at least within the tank. Therefore, the
Transformer and Reactor Protection 207

Figure 10-50 Interconnected delta zig-zag transformer with voltages 608 apart on the two sides.

three 50/51 relays represent backup protection, which 10.5 Reactors on Delta System
could be omitted.
The type SA-1 or CA generator differential relays or On delta systems, shunt reactors are usually con-
bus differential type KAB relays can also be used for nected to the tertiary of a power transformer
the ground differential (87N). Additional security can associated with the line. Since most faults will involve
be obtained by using the type CWC relay, particularly ground, the units or associated system are grounded
if the current transformer performance is inferior. This through high resistance for detection purposes. Neu-
connection is shown in Figure 10-36. Also, the ratios tral resistance grounding is shown in Figure 10-56,
do not have to be identical. Sensitivity can be increased and voltage transformer grounding in Figure 10-57
by using a lower-ratio neutral current transformer, as (see Chap. 7). To limit both transient overvoltage and
described above. ground fault current, the resistor is sized so that I0R
When the shunt reactor is grounded and connected equals or exceeds I0C. Since the system capacitance to
to an ungrounded system, a CWC relay can be used, ground is very large, the impedance of the associated
with the connections shown in Figure 10-37. system is essentially negligible and is not shown in the
208 Chapter 10

Figure 10-51 Interconnected wye zig-zag transformer with wye phase voltages lagging 308 from zig-zag phase voltage.

zero sequence diagrams. Whereas the primary current Phase protection for three-phase reactors can be
for a ground fault is quite small, the secondary obtained by overcurrent or differential relay schemes.
current will be large. If 59N is used for alarm instead Overcurrent protection is the same as for Figure 10-52,
of tripping, the secondary current may exceed the without 50N/51N; differential protection is as shown
continuous thermal rating of the voltage transfor- in Figure 10-54 or 10-53, without 50N/51N. The
mers. arrangements offer little protection for single-phase
The CV-8 relay for 59N provides sensitive protec- reactors unless a second ground fault should develop in
tion: Its pickup is 8% of its continuous rating. For another unit.
alarm applications, the 3E0 voltage should not exceed Although including the reactor within the transfor-
the 69- or 199-V rating unless a series resistor is used to mer differential circuit provides some phase-fault
limit the voltage across the relay to its rating. (See protection, it offers no ground-fault protection with
Chap. 8.) high impedance grounding. Even the phase-fault
Transformer and Reactor Protection 209

Figure 10-52 Phase and ground instantaneous and time

overcurrent protection for shunt reactors.

Figure 10-54 Phase instantaneous and time overcurrent

with ground differential protection for shunt reactors.

offers the greatest sensitivity, its application is limited.

The reactors must be oil-type, and the fault must cause
a sufficient pressure change to operate the unit. While
transformer action in a turn-to-turn fault can produce
a large current within the shorted turn, there is very

Figure 10-53 Separate phase differential protection with

ground time overcurrent backup protection for shunt

protection is limited, since the current transformers of

the transformer differential are sized for transformer
capacity and not the smaller-reactor MVA.
Low impedance or solid grounding of the reactors
may be used. In this case, either the 50N/51N neutral
overcurrent relay (Figs. 10-56 and 10-57) or 87N
ground differential of Figure 10-37 should be applied.

10.6 Turn-to-Turn Faults

Light turn-to-turn faults are extremely difficult to Figure 10-55 Combined phase and ground differential
detect. Although the rate-of-rise-of-pressure relay protection for shunt reactors.
210 Chapter 10

Figure 10-56 Neutral resistance grounding for ground fault detection.

Figure 10-57 Resistance grounding through voltage transformer for ground detection.
Transformer and Reactor Protection 211

little current change at the terminals of the unit. The

effect is equivalent to an autotransformer with a
shorted secondary. The impedance change that will
occur in one phase can be represented by symmetrical
components as a shunt unbalance. As shown in
Figure 10-58, impedance ZA of phase A is not equal
to the other two phases, shown with a total reactor
impedance of ZB. For this condition, the sequence
networks are connected as shown in Figure 10-58.
Because of the transformer action, the change in
impedance of the total phase circuit for a shorted turn
is difficult to calculate. As a rough estimate, assume a
change of 3% in the phase with the shorted turn. Also Figure 10-58 Shunt reactor with a shortened turn in phase
assume that the fault has not yet involved ground or ‘‘a’’ so that ZA < ZB and the sequence connections for an
unbalanced impedance in phase ‘‘a.’’
other phases. Given these assumptions and if we
neglect phase angles, distributed winding capacitance,
and transformer action, negative and zero sequence conditions. Although special schemes or relays have
currents will be less than 1%. Removing the ground been reported, they will require very careful ‘‘custo-
from the units does not change the positive and mized’’ applications.
negative sequence currents significantly, although it As the turn-to-turn fault spreads to more turns, the
does eliminate the zero sequence. The magnitudes of current will increase. A negative sequence relay may be
the currents are largely a function of the total reactor set at 0.2 A negative sequence. The relay should be
impedance; the source impedance is relatively low applied with a timer to avoid operation on system
compared to the reactor impedance. transients and external faults and should be disabled
The small unbalances and sequence currents asso- when the breaker is opened. This latter safeguard
ciated with turn-to-turn faults generally are no larger avoids possible operation on low-frequency line
than the normal or tolerable unbalances. Conse- oscillations after the line is deenergized. With very
quently, there seems to be no reliable ‘‘handle’’ to little resistance in the line, such oscillations can last an
distinguish between the intolerable and tolerable appreciable time.
Station-Bus Protection
Revised by: SOLVEIG WARD

1 INTRODUCTION 1.1 Current Transformer Saturation Problem

and Its Solutions on Bus Protection
A bus is a critical element of a power system, as it is the
point of convergence of many circuits, transmission, Bus differential relaying is complicated by the fact that
generation, or loads. The effect of a single bus fault is for an external fault on one circuit, all of the other
equivalent to many simultaneous faults and usually, circuits connected to the bus contribute to that fault.
due to the concentration of supply circuits, involves The current through the circuit breaker for the faulted
high-current magnitudes. High-speed bus protection is circuit will be substantially higher than that for any of
often required to limit the damaging effects on the other circuits. With this very high current flowing
equipment and system stability or to maintain service through the current transformer and its circuit breaker,
to as much load as possible. The bus protection there is a very high likelihood that some degree of
described refers to protection at the bus location, saturation will occur. A saturated current transformer
independent of equipment at remote locations. will not deliver its appropriate current to the bus relay.
Differential protection is the most sensitive and With the lower currents in the other circuits for this
reliable method for protecting a station bus. The external fault, the degree of saturation is expected to be
phasor summation of all the measured current entering considerably lower. This may lead to a large differ-
and leaving the bus must be 0 unless there is a fault ential current that will tend to cause the relay to sense
within the protective zone. For a fault not in the an internal fault rather than the actual external fault
protective zone, the instantaneous direction of at least that exists. The relay must accommodate this error
one current is opposite to the others, and the sum of current without misoperation.
the currents in is identical to the sum out. A fault on A widely used equivalent diagram for a current
the bus provides a path for current flow that is not transformer appears in Figure 11-3b. It consists of a
included in these summations. This is called the perfect transformation from the high current side to
differential current. Detection of a difference exceeding the low current side (e.g., 600:5). All of the significant
the predictable errors in the comparison is one imperfections are lumped into Rp, Rs, and Xm. The Rs
important basis for bus relaying. represents the internal secondary resistance of the ct
In dealing with high-voltage power systems, the (current transformer), and the X represents a current
relay is dependent on the current transformers in the path that accommodates the exciting requirements.
individual circuits to provide information to it regard- The ct is assumed to have a uniformly distributed
ing the high-voltage currents. Figure 11-1 shows winding and, therefore, to manifest no significant
typical examples of the location of current transfor- leakage reactance.
mers that are used for this purpose. The arrowheads When the ct is subjected to excessive flux, the ct is
indicate the reference direction of the currents. said to ‘‘saturate,’’ meaning that the core of the ct has


Figure 11-1 Common bus arrangements with relay input sources.

Chapter 11
Station-Bus Protection 215

been forced to carry more flux than it can handle. The (zone-2 trip) and may interrupt unnecessarily a tapped
flux then spills into the area surrounding the core, load. When local bus protection is applied, the
causing the magnetizing reactance to have a much following information is required for the scheme
lower value than normal. It can be seen that any selection, relay selection, and setting calculations:
current that flows in X subtracts from the perfectly
1. Information about the bus configuration is
transformed current, producing a deficiency in the
required. The common bus arrangements are
current that is delivered to the devices connected to the
as shown in Figure 11-1, such as single bus,
ct. The black blocks are the polarity markers. A single
double bus, main-and-transfer bus, ring bus,
polarity marker has no significance. With two, it is
breaker and a half, bus tie-breaker, double-bus-
acknowledged that, at the instant of time at which
single-breaker, etc.
current is flowing into the polarity marker on the high
2. Maximum and minimum bus fault currents
current side of the ct, current is flowing out of the
(single-phase-to-ground fault and three-phase
polarity marker on the low current side. Of course, the
current reverses every half cycle, but both the high and
3. Current transformer information, including
low reverse together.
Direct current saturation is much more serious than Current transformer location
ac saturation because a relatively small amount of dc Current transformer ratios
from an asymmetrical fault wave will saturate the Current transformer accuracy class
current transformer core and appreciably reduce the Current transformer saturation curves
secondary output. The L/R ratio of the power-system
4. Operating speed requirement
impedance, which determines the decay of the dc
component of fault current, should strongly influence
the selection of the bus protective relaying. Typically,
1.3 Normal Practices on Bus Protection
the dc time constants for the different circuit elements
can vary from 0.01 sec for lines to 0.3 sec or more for
The normal practices on bus protection are
generating plants. The nearer a bus location is to a
strong source of generation, the greater the L/R ratio 1. There is one set of bus relays per bus section.
and the slower the decay of the resulting dc component 2. Use a dedicated ct for bus differential protec-
of fault current. tion. If possible, the connection of meters,
Of the several available methods for solving the auxiliary ct’s, and other relays in differential-
unequal performance of current transformers, four are type bus schemes should be avoided since these
in common use: devices introduce an additional burden into the
main circuit.
1. Eliminating the problem by eliminating iron in
3. Lead resistance, as well as ct winding resistance,
the current transformer [a linear coupler (LC)
contributes to ct saturation. Therefore, the
length of secondary lead runs should be held
2. Using a multirestraint, variable-percentage dif-
to a minimum.
ferential relay which is specifically designed to
4. Usually, the full-ct secondary winding tap
be insensitive to dc saturation (CA-16 relay
should be used. This has two advantages. It
minimizes the burden effect of the cable and,
3. Using a high impedance differential relay with a
second, leads by minimizing the secondary
series resonant circuit to limit sensitivity to ct
current and makes use of the full-voltage
saturation (KAB relay system)
capability of the ct.
4. Using a Differential Comparator relay with
5. Normally, there is no bus relay required for the
moderately high impedance to limit sensitivity
transfer bus on a main-and-transfer bus
to ct saturation (RED-521)
arrangement. The transfer bus is normally
deenergized and will be included in the main
1.2 Information Required for the Preparation of bus section when it is energized.
a Bus Protective Scheme 6. No bus relay is required for a ring bus because
the bus section between each pair of circuit
Some bus protection schemes rely on the operation of breakers is protected as a part of the connected
a remote breaker. It is simple and economic, but slow circuit.
216 Chapter 11

7. Special arrangements should be considered if than three primary time constants, and its setting must
there is any other apparatus, such as station be greater than the exciting current of the current
service transformers, capacitor banks, ground- transformer under worst-case conditions. This may
ing transformers, or surge arresters, inside the require a setting of 10 or more amperes and a time
bus differential zone. setting of, say, 18 cycles. These values may be
8. There is no simple scheme available for a acceptable for smaller substation buses, but more
double-bus-single-breaker arrangement (Fig. sophistication and faster relaying speed are generally
11-1e), because its current transformers are mandatory for more extensive and higher-voltage
normally located on the line side. These buses.
applications greatly benefit from numerical In these applications a ‘‘short time’’ or ‘‘extremely
schemes, such as the RED-521. (Refer to Sec. inverse’’ characteristic overcurrent relay is used in the
9 of this chapter for more information.) interests of getting faster tripping speeds at high
current. Operating times of 8 to 18 cycles are expected.
Although the relay cost is low, the engineering cost
may be high because of the usual need for considerable
2 BUS DIFFERENTIAL RELAYING WITH study for the application to assure correct operation.
2.1 Overcurrent Differential Protection
2.2 Improved Overcurrent Differential
This differential scheme requires that a time-over- Protection
current relay be paralleled with all of the current
transformers for a particular phase, as shown in The sensitivity of the overcurrent differential scheme
Figure 11-2. It is permissible to use auxiliary ct’s to (Fig. 11-2) can be improved by externally connecting a
match ratios, but it is preferred that all of the ct’s have series resistor with each overcurrent relay, as shown in
the same ratio on the tap chosen and that the use of Figure 11-3. These resistors are called stabilizing
auxiliary ct’s be avoided. resistors. If we assume that an external fault causes
In this scheme, the overcurrent relay must be set to the ct on the faulted feeder to be saturated completely,
override the maximum error current that results from the ct excitation reactance will approach 0. As shown
an external fault (phase or ground). It may also be in Figure 11-3, the error current Id that flows through
necessary to have sufficient time delay to refrain from
tripping during the time that one or more of the
current transformers is severely saturated by the dc
component of the primary current. To assure this,
using a simple overcurrent relay, the current transfor-
mers must be chosen to have no more than 20 times
rated current flowing in their primary for the worst-
case external fault, and each have a burden no more
than the rated value (relaying-accuracy-class voltage/
100). The operating time of the relay must not be less

Figure 11-3 The improved overcurrent differential bus

Figure 11-2 The overcurrent differential bus protection. protection.
Station-Bus Protection 217

the overcurrent unit would be percentage is small for maximum sensitivity. For heavy
  external faults, current transformer performance is
2RL þ RS
Id ¼ IF ð11-1Þ likely to be poor, and the percentage is large. The
2RL þ RS þ Rd variable-percentage characteristic is obtained by ener-
where Rd is the resistance in the differential path. gizing the operating unit through a built-in saturating
In order to reduce the error current Id in the autotransformer.
differential path for improving the sensitivity of the The saturating autotransformer also presents a high
scheme, the most effective way is to increase the value impedance to the false differential current, which tends
of Rd. The limitations of this additional resistance are to limit the current through the operating coil and to
determined by (1) the overvoltage to the ct circuit and force more equal saturation of the current transfor-
(2) the minimum available internal fault current. It mers. On internal faults, in which a desirable high
should be limited to differential current exists, saturation reduces the
impedance. A further advantage of the saturating
VCL autotransformer is that it provides a very effective
Rd ¼ O ð11-2Þ
46Imin pickup shunt for the dc component, appreciably reducing the
dc sensitivity of the operating units. At the minimum
Note: The multiplier 4 includes a safety factor of 2.
pickup current of 0.15 + 5% A, the restraining coils
are ineffective.
When using the CA-16 relay, the current transfor-
3 MULTIRESTRAINT DIFFERENTIAL SYSTEM mers should not saturate when carrying the maximum
external symmetrical fault current; that is, the exciting
Multirestraint differential schemes use conventional current should not exceed one secondary ampere rms.
current transformers, which may saturate on heavy This requirement is met if the burden impedance does
external faults. For this reason, the secondary current not exceed
output may not represent the primary. In a differential
scheme, the current transformers and relay function as ½NP VCL  ðIEXT  100ÞRS
a team. When the current transformers do not perform 1:33 IEXT
adequately, the relay can within limits make up for the
The multirestraint differential scheme uses the CA- NP ¼ proportion of total current transformer
16 variable-percentage differential relay, which con- turns in use
sists of three induction restraint units and one VCL ¼ current transformer accuracy-class voltage
induction operating unit per phase. Two of the units IEXT ¼ maximum external symmetrical fault current
are placed opposite each other and operate on a in secondary (amperes rms) (use IEXT ¼ 100
common disc. In turn, the two discs are connected to a if IEXT < 100)
common shaft with the moving contacts. All four of RS ¼ current transformer secondary winding resis-
the units are unidirectional; that is, current flow in tance of the turns in use (in ohms); for
either direction through the windings generates con- example, if the 400:5 tap of a 600:5 wye-
tact-opening torque for the restraint units or contact- connected class C200 current transformer is
closing torque for the operating unit. Each restraint used, then NP ¼ 400/600 ¼ 0.67 and
unit (called R, S, and T) also has two windings to VCL ¼ 200
provide restraint proportional to the sum or difference,
depending on the direction of the current flow. If the If IEXT ¼ 120 A and RS ¼ 0.5 O, then the burden of
currents in the two paired windings are equal and the ct’s secondary circuit, excluding current transfor-
opposite, the restraint is cancelled. Thus, the paired mer secondary winding resistance, should not exceed
restraint windings have a polarity with respect to each 0:676200  ð120  100Þ0:5
other. With this method six restraint windings are ¼ 0:78 O
available per phase.
In addition to providing multiple restraint, the Settings for the CA-16 relay need not be calculated.
variable-percentage characteristic helps in overcoming Field experience indicates that one CA-16 relay per
current transformer errors. At light fault currents, phase is satisfactory for the vast majority of applica-
current transformer performance is good, and the tions.
218 Chapter 11

External connections are as shown in Figures 11-4

through 11-6. Figure 11-4 may be used if only three
circuits are involved. The term circuit refers to a source
or feeder group.
When several circuits exist and the bus can be
reduced to four circuits, then the scheme of Figure 11-5
may be used. For example, assume a bus consists of
two sources and six feeders, and that the feeders are
lumped into two groups. The bus now reduces to four
In paralleling current transformers, each feeder
group must have less than 14 A load current (restraint
coil continuous rating).
If the bus reduces to more than four circuits, then
the scheme of Figure 11-6 should be used. In applying
the scheme of Figure 11-6, each primary circuit must
be identified as either a source or feeder. As defined
here, a feeder contributes only a small portion of the
total fault current for a bus fault. All other circuits are
sources. Next, a number of feeders are lumped into a
Figure 11-5 Connection of one CA-16 relay per phase to
feeder group by paralleling feeder current transfor- protect a bus with four equivalent circuits. (Connections for
mers. Each feeder group must have less than 14 A load one phase only are shown.)
current and not contribute more than 10% of the total
phase- or ground-fault current for a bus fault. Then
connect the ‘‘source’’ and ‘‘feeder groups’’ alternately
as shown in Figure 11-6.
Note that in Figures 11-4 through 11-6, electro-
magnets R, S, and T are referred to. Each of these
elements has two windings. The polarity markings are
extremely significant as related to one another on the
same electromagnet, but have no significance with
respect to one another on different electromagnets. If
the current into a polarity marker is equal to the

Figure 11-4 Connection of one CA-16 relay per phase to Figure 11-6 Connection of one CA-16 relay per phase to
protect a bus with three equivalent circuits. (Connections for protect a bus with six equivalent circuits. (Connections for
one phase only are shown.) one phase only are shown.)
Station-Bus Protection 219

current out of the polarity marker on the same overcurrent differential’’ approach. It uses a high
electromagnet, there will be no restraining torque impedance voltage element instead of ‘‘a low impe-
produced by that electromagnet. The sum of all of the dance overcurrent element plus an external resistor.’’
restraint torques is compared to that produced by the The high impedance differential KAB relay consists
operating coil. Current into the operating coil circuit of an instantaneous overvoltage cylinder unit (V), a
produces a much stronger effect than the same current voltage-limiting suppressor (varistor), an adjustable
through a single restraint winding. For an external tuned circuit, and an instantaneous current unit (IT).
fault, there is no current through the operating coil if On external faults, the voltage across the relay
the current transformers perform perfectly. There will terminals will be low, essentially 0, unless the current
be substantial restraint for this same condition, even transformers are unequally saturated. On internal
though in some restraint electromagnets some (or even faults, the voltage across the relay terminals will be
total) cancellation may take place. high and will operate the overvoltage unit. Since the
Consider a fault on the bus of Figure 11-5 in which impedance of the overvoltage unit is 2600 O, this high
all of the high-voltage circuits contribute the same voltage may approach the open-circuit voltage of the
value of current. All of the restraint cancels because in current transformer secondaries. The varistor limits
each of the electromagnets the current into the polarity this voltage to a safe level.
marker equals the current out of its paired coil. All of Since offset fault current or residual magnetism
the internal fault current (in secondary terms, of exists in the current transformer core, there is an
course) flows into the operating coil circuit and fast appreciable dc component in the secondary current.
tripping occurs. Practical cases with widely differing The dc voltage that appears across the relay will be
fault contributions produce similar effects even though filtered out by the tuned circuit, preventing relay
considerable restraint torque may be present. pickup.
Consider, now, an external fault on the upper circuit The IT current unit provides faster operation on
off of the bus with the equal fault current contributions severe internal faults and also backup to the voltage
that were assumed in the previous case. Torque unit. The range of adjustment is 3 to 48 A.
cancellation occurs in electromagnet T, as before. The KAB relay has successfully performed opera-
Substantial restraint torque is produced by R and S. tions up to external fault currents of 200 A secondary
The operating coil current cannot exceed the error and down to an internal fault current of 0.27 A
current in the faulted circuit (which may well be secondary. Its typical operating speed is 25 msec.
extreme due to the effect of saturation). The overvoltage unit is set by calculating the
This is a very sensitive bus relaying scheme, and it is maximum possible voltage for an external fault as
very secure against operation for external faults even follows:
though severe ct saturation may occur for one or more
ct’s. It is reasonably fast. Another advantage is that it IF
VR ¼ KðRS þ RL Þ ð11-4Þ
can accept auxiliary ct’s in the circuit, which allows N
different ratios of the main ct’s. Two shortcomings are
its comparative inflexibility as other circuits are added
to the bus and the need to bring all circuits back from VR ¼ pickup setting of the V unit in volts rms
the switchyard to the relay location. RS ¼ dc resistance of current transformer secondary
winding, including internal leads to bushing
4 HIGH IMPEDANCE DIFFERENTIAL SYSTEM RL ¼ resistance of lead from junction points to the
most distant current transformer (one-way
Although the high impedance differential scheme also lead for phase faults, two-way lead for
uses conventional current transformers, it avoids the phase-to-ground faults)
problem of unequal current transformer performance IF ¼ maximum external primary fault current, in
by loading them with a high impedance relay (Fig. amperes rms, contributed by the bus
11-7). N ¼ current transformer turns ratio
This arrangement tends to force the false differential K ¼ margin factor
currents through the current transformers rather than
the relay operating coil. Actually, the high impedance The maximum voltage occurs for the external fault
differential concept comes from the above ‘‘improved when the faulted circuit current transformer is
220 Chapter 11

Figure 11-7 External connection of type KAB bus differential relay.

completely saturated, and there is no saturation in the KAB relay application, the knee voltage is
source current transformers. The maximum voltage is defined as the intersection of the extension of
equal to the resistance drop produced by the secondary the two straight-line portions of the saturation
current through the leads and secondary winding of curve. The ordinate and abscissa must use the
the saturated current transformer. In practice, the same scales.
faulted current transformer will never completely
saturate, and the source current transformers will The margin factor curve, shown in Figure 11-8, is
tend to saturate. As a result, the actual maximum based on tests of the KAB relay in the high-power
voltage is less than the theoretical value. The margin laboratory. A safety factor of 2 has been included in
factor K, which modifies this voltage, varies directly constructing this curve.
with the current transformer saturation factor SF: The maximum number of circuits that can be
connected to the relay, or the minimum internal fault
1 ðRS þ RL ÞIF current required to operate the relay, can be estimated
¼ ð11-5Þ
SF NVk from the following equation:
Imin ¼ ðXIe þ IR þ IV ÞN ð11-6Þ
Vk ¼ knee voltage value of the poorest current
transformer connected to the relay. For type where
Station-Bus Protection 221

The use of auxiliary ct’s is discouraged, though,

with proper consideration of their resistance in
series with the lead resistance (raising the
effective RL), they may be used at the sacrifice
of some sensitivity of fault recognition. The same
comment applies to the introduction of other
devices in the current transformer circuits.
The junction point for all of the ct’s in the bus
differential system should be in such a location as
to equalize as much as possible the distance from
each ct to this point. This will minimize RL, the
value used in the setting calculation and thus
allow better sensitivity to be achieved. Departure
from this requirement is permissible in metal-clad
Figure 11-8 Empirical margin factor for setting the V-unit switchgear because of the comparatively short
of the KAB relay. distances usually involved.
The lead resistance from the junction point to the
Imin ¼ minimum primary fault current in amperes relay terminals is not critical.
Note that with this system total saturation of the
Ie ¼ secondary excitation current of the current
current transformer on a circuit feeding an external
transformer at a voltage equal to the setting
fault is allowed and the relay remains secure.
value of the V unit in amperes
IR ¼ current in the V unit at setting voltage VR in
amperes, that is, IR ¼ VR/2600 4.2 Factors that Relate to the High-Voltage
IV ¼ current in varistor circuit at a voltage equal to Problem
the setting value of the V unit in amperes
(generally negligible) All ct’s in the bus differential circuit should be
N ¼ current transformer turns ratio operated on their full-tap position. Refer to
X ¼ number of circuits connected to the bus Figure 11-9; a high voltage will be induced on the
In general, the following factors should be considered unused portion of the ct circuit due to auto-
when applying a high impedance bus differential relay. transformer action.
All current transformers should have the same ratio.
If taps must be used, the windings between the
4.1 Factors that Relate to the Relay Setting

The V-unit setting of the KAB relay is based on the

calculated result of Eq. (11–4), which is determined by
the values of K, RS, RL, and IF. In order to keep this
setting value within the available relay range of 75 to
400 V, it is necessary to keep the values of (RS þ RL)
and any additional burden in the ct secondary as low
as possible. This includes the consideration of the
Use fully distributed winding current transformers,
such as bushing ct’s or current transformers with
toroidally wound cores, such as those used in
metal-clad switchgear. These ct’s provide a
negligible leakage reactance and therefore do
not contribute to the internal impedance in the
equivalent circuit of the ct. Only the RS resistance Figure 11-9 High voltage induced by autotransformer
is needed in series with RL in Eq. (11–4). action.
222 Chapter 11

taps must be completely distributed, and any

high voltage at the full-tap terminal caused by
autotransformer action should be checked to
avoid insulation breakdown. In general, auxiliary
ct’s should not be used to match ratios.

4.3 Setting Example for the KAB Bus Protection

Assume a six-circuit bus for which the maximum

external three-phase fault current is 60,000 A rms,
symmetrical; the maximum external phase-to-ground
fault current is 45,000 A, and the minimum internal
fault current is 10,000 A. The current transformer
ratios are 2000:5, ANSI class C400, Vk is 375 V. The
secondary winding resistance RS is 0.93, and one-way
lead resistance to junction point RL is 1.07 O. Figure 11-10 Setting of KAB instantaneous unit.

the example, the ordinate value is

4.3.1 Settings for the V Voltage Unit ð0:93 þ 1:07Þ60,000
Three-phase fault ¼ ¼ 300
For the three-phase fault condition [using Eq. (11-5)], 400
Phase-to-ground fault
1 ð0:93 þ 1:07Þ60,000
¼ ¼ 0:8 ð0:93 þ 261:07Þ45,000
SF 4006375 ¼ ¼ 345
From Figure 11-8, 1.2 > K  0.82 (use the lower value
of 0.82 for sensitivity); therefore, using Eq. (11-4), we From these numbers, it is obvious from Figure 11-10
get that the IT unit is incapable of operating for an
external fault. The lowest available setting of 3 A will
60,000 usually be adequate because of the high conduction
VR  0:82ð0:93 þ 1:07Þ ¼ 246 V
400 level of present-day varistors. The principal tripping
For the phase-to-ground fault condition, function is accomplished at high speed by the voltage
unit, and only in extreme circumstances will the IT unit
1 ð0:93 þ 261:07Þ645,000 operate for an internal fault.
¼ ¼ 0:92
SF 4006375
And from Figure 11-8, 1.1 > K  0.77; therefore, using
VR  0:77ð0:93 þ 261:07Þ ¼ 266 V These relays use the fundamental principle described in
400 Figure 11-11. The RADSS is a solid-state version, the
The minimum setting of the V unit in the KAB relay, REB-103 is similar to this, but the logic is accomplished
therefore, should be 266 V, the larger value for either with a microprocessor, while the RED-521 is entirely a
the three-phase or phase-to-ground conditions, as numerical relay. All are very high-speed relays (9- to 16-
calculated. msec tripping) and are very secure against misoperation
for external faults; all reliably and sensitively detect
internal faults and are quite flexible in accommodating
additional circuits. They may also be used for generator
4.3.2 Setting for the IT Current Unit
stator protection and for shunt reactor protection
The IT setting is determined from Figure 11-10. The though their prime application area is for bus protection.
higher value is used as the ordinate as determined from The RADSS and the REB-103 relays use external
the three-phase and phase-to-ground fault. Thus, for auxiliary current transformers which allow substan-
Station-Bus Protection 223

Figure 11-11 Differential comparator relay.

tially different main circuit current transformers to be associated with the circuit on which the external fault
accommodated and also reduce current to a suitable occurred, restraint is developed across the resistor Rs.
level for the relay. The RED-521, being a micropro- Any condition that produces I DIFF current will,
cessor relay, is able to accept widely varying inputs through the transformer and the full-wave bridge,
from the main current transformers and to provide, generate a voltage Vd3. For the through fault case, the
internally, the appropriate scaling factors. The RED- restraint voltage Vs will exceed the operating voltage
521 is therefore very suitable for double-bus-single- Vd3, and the relay will refrain from operating. For the
breaker arrangements as no external ct switching takes internal fault case, I DIFF will be large, Vd3 will exceed
place. The ct is connected to the appropriate protection Vs, current will be passed through the diode and the
zone numerically inside the relay. reed relay DR, and tripping will occur. SR is a ‘‘start’’
Taking advantage of Kirchoff’s law, the scheme relay whose contact supervises tripping to add to the
compares the sum of all of the currents entering the overall security of the relay. It is obvious that this relay
bus with the sum of all of the currents leaving the bus. is extremely fast because the decision to trip is based
These are instantaneous currents (as opposed to rms or on instantaneous currents.
average currents.) The RED-521 numerical relay uses this principle,
In the circuit of Figure 11-11, the currents are but is not encumbered by need for the auxiliary
delivered to the relay through the diodes. The sum of matching current transformers, the diodes, or any
the currents through the lower group of diodes is other of the components inherently required in the
representative of the instantaneous sum of the incom- comparison process. The individual samples of cur-
ing currents to the bus, and the current flowing to the rents are collected and summed appropriately to
upper group of diodes is representative of the develop numerically the I IN and I OUT values and
instantaneous sum of the currents leaving the bus. the corresponding restraint quantity. This is compared
These two sum currents are always in perfect balance with the difference of these individual sums, I DIFF,
provided the current transformers perform their job and a determination of the need to trip is established.
faithfully and there is no fault on the bus (or to state it
more correctly, provided there are no current paths off
of the bus that are unaccounted for). 6 PROTECTING A BUS THAT INCLUDES A
If an internal fault (phase or ground) were to occur, TRANSFORMER BANK
the currents in and out would no longer match. They
would differ by the amount of the fault current. This Ideally, when the bus includes a power transformer
difference current appears as I DIFF in the relay. bank, separate protection should be provided for the
To accommodate the inherent errors in the current bus and transformer, even though both protection
transformers for an external fault, particularly in the ct schemes must trip all breakers around the two units.
224 Chapter 11

Such a system offers maximum continuity of service,

since faults are easier to locate and isolate. Also, using
a bus differential relay for bus protection and
transformer differential relay for transformer protec-
tion provides maximum sensitivity and security with
minimum application engineering.
However, economics and location of current trans-
formers often dictate that both units be protected in
one differential zone. For these applications, either the
multirestraint HU-4 or CA-26 relays should be used.
The HU-4 relay is similar to the HU and HU-1 relays,
except that it has four restraint windings. Also, the Figure 11-13 Protection of a typical transformer section
rectified outputs of the restraint transformers are where the transformer tertiary is brought out for load or
connected in series, providing a higher restraint force connected to an external source.
when a through fault occurs on the bus. Since the dc
saturation of current transformers will allow current to
pass into the HRU transformers and possibly pick up the The CA-26 relay is applicable to relatively small
IIT, the IIT unit of the HU-4 relay is set at 15 times the transformers remote from generating stations, HV,
rms tap value to prevent false tripping for external faults. and EHV buses. Here, inrush will usually be light and
Similar to the CA-16, the CA-26 relay has a not cause the CA-26 to operate. If, however, complete
stronger contact spring and higher pickup of security against inrush is required, the HU-4 must be
1.25 + 5% A to help override inrush. Its variable applied.
restraint curve is steeper than the CA-16, and its With CA-26 relays, the four-circuit bus connections
operating time is approximately three cycles. of Figure 11-5 are not recommended for bus protec-
Of the two types, the HU-4 relay is preferred, as it is tion, since the relay may have too much restraint for a
immune from operation on transformer magnetizing bus fault.
inrush. The HU-4 should always be applied for large The bus CA-16 relay should not be used for the
transformer banks or those associated with HV and transformer differential, since it is too sensitive to
EHV buses. A typical application, shown in Figure 11- override magnetizing inrush.
12, protects a three-winding transformer bus with four
circuits. Figure 11-13 illustrates another typical appli-
cation used in EHV systems. 7 PROTECTING A DOUBLE-BUS SINGLE-

The double-bus single-breaker with bus tie (Figure 11-

1e) provides economic and operating flexibility com-
parable to the double-bus double-breaker arrangement
(Fig. 11-1c). However, the ct’s are normally on the
line-side location, which results in increased differen-
tial relaying problems. Two different approaches have
been used in the bus protection of such arrangements:
the fully switched scheme (Fig. 11-14) and the
paralleling switch scheme (Fig. 11-15). They are both
complicated (inserting switch contacts in the ct
circuits) and/or imperfect in protection. These schemes
either require switching ct’s and/or disabling the bus
protection before any switching operation. This is a
period when the probability of a bus fault occurring is
Figure 11-12 Typical application of HU-4 relay for high and it is most desirable that the bus protection be
protecting a large transformer bank associated with HV in service. A third scheme as shown in Figure 11-16 can
and EHV buses. (Auxiliary current transformers for ratio be considered. It is similar to the paralleling switched
matching are not shown.) scheme except a check-zone relay is added as shown.
Station-Bus Protection 225

Figure 11-14 Fully switched scheme.

Figure 11-15 Paralleled switched scheme.

226 Chapter 11

Figure 11-16 Paralleled switch with check zone scheme.

Two bus differential zones are provided, one for each in one or more of the circuits are not included in the
bus, with each one overlapping the bus breaker. Each phasor summation of the current to the relay.
primary circuit is normally switched to a specific bus, In this scheme, only the source circuits are
and relay input circuits and breaker control circuits are differentially connected, as shown in Figure 11-17b,
wired accordingly. The additional check-zone device using a high-set overcurrent relay with time delay. The
supervises the trip circuits. If it becomes necessary to ct’s protecting the feeders or circuits are not in the
clear one of the buses, all the primary circuits may be differential connection.
switched to the opposite bus and it is needless to Essentially, this arrangement combines time-delay
disable the bus protection before any switching bus protection with feeder backup protection. The
operation. However, this scheme still has two draw- sensitivity and speed of this scheme are not as good as
backs when any one or all of the primary circuits is with complete differential protection. This method
switched to the opposite bus: (1) It will lose its may be used as a backup to a complete differential
selectivity, and (2) it will reduce its sensitivity since scheme, as primary protection for a station with loads
the two relays are paralleled. protected by fuses, or to provide local breaker failure
A numerical scheme, such as RED-521, overcomes protection for load breakers.
these drawbacks as there is no external ct switching In modern microprocessor systems, provision has
involved. The ct’s are connected to the appropriate been included to allow communications between the
zone by numerical switching in the relay. feeder breaker relaying and the source breaker relaying.
The feeder breakers are each equipped with a nontrip-


Other methods for protecting buses are in limited use:

(1) partial differential schemes, (2) directional compar-
ison relaying, and (3) the fault-bus method. Except for
the latter, these schemes are most often applied as
economic compromises for the protection of buses.

8.1 Partial Differential Relaying

This type of protection is also referred to as ‘‘bus

overload’’ or ‘‘selective backup’’ protection. It is a
variation of the differential principle in which currents Figure 11-17 Partial differential protection.
Station-Bus Protection 227

ping low-set instantaneous overcurrent function that is

set somewhat above their maximum load. The source
breakers have an instantaneous overcurrent unit with
slight time delay that is set above the maximum total
load current for the bus, and they are equipped to
receive a status input from the feeder breakers. For a
fault on one of the feeder circuits, the low-set
instantaneous overcurrent unit operates and applies a
block signal to the source relay. The instantaneous unit
of the source breaker operates, but is unable to trip
because of the block signal. The time-delayed and
coordinated tripping of the source breaker is not
affected so its backup function stays intact.
For a bus fault, the block signal is absent, and
tripping of the source breaker occurs at high speed.
Some partial differential circuits use distance-type
relays in the scheme. The use of a distance relay for this
scheme produces both faster and more sensitive
operation than the overcurrent scheme.

8.2 Directional Comparison Relaying

Figure 11-18 Directional comparison bus protection.

Occasionally, it is desirable to add bus protection to an

older substation where additional ct’s and control cable and have only one connection to ground. An over-
are too costly to install. In this instance, the existing ct current relay is connected in this ground path as shown
circuits used for line relaying can also be used for the in Figure 11-19. Any ground fault to the supporting
directional comparison bus relaying protection. structure will cause fault current to flow through the
As shown in Figure 11-18, the directional comparison relay circuit, tripping the bus through the multiple-
relaying uses individual directional overcurrent relays contact auxiliary tripping relay. A fault detector,
on all sources and instantaneous overcurrent relays on energized from the neutral of the grounded transfor-
all feeders. The directional relays close contacts when mer or generator, prevents accidental tripping. This
fault power flows into the bus section. Back contacts on scheme requires special construction measures and is
the overcurrent relays open when the fault is external on expensive.
the feeder. All contacts are connected in series, and when
the fault occurs on the bus, the trip circuit is energized
through a timer. A time delay of at least four cycles will
allow all the relays to decide correctly the direction of the
fault and to permit contact coordination.
In this scheme, the ct’s in each circuit do not require
the same ratio and can be used for other forms of
relaying and metering.
The disadvantage of this scheme is the large number
of contacts and complex connections required. There is
also the remote possibility of the directional elements
not operating on a solid three-phase bus fault as a
result of 0 voltage.

8.3 Fault Bus (Ground-Fault Protection Only)

This method requires that all the bus supporting

structure and associated equipment be interconnected Figure 11-19 Fault bus.
Line and Circuit Protection
Revised by: ELMO PRICE

1 INTRODUCTION 1. Instantaneous overcurrent

2. Time overcurrent
1.1 Classification of Electric Power Lines
3. Directional instantaneous and/or time over-
Alternating current lines are commonly classified by
4. Step time overcurrent
function, which is related to voltage level. Although
5. Inverse time distance
there are no utility-wide standards, typical classifica-
6. Zone distance
tions are as follows:
7. Pilot relaying
1. Distribution (2.4 to 34.5 kV) Circuits trans-
mitting power to the final users.
2. Subtransmission (13.8 to 138 kV) Circuits
1.3 Selecting a Protective System
transmitting power to distribution substations
and to bulk loads.
Several fundamental factors influence the final choice
3. Transmission (69 to 765 kV) Circuits trans-
of the protection applied to a power line:
mitting power between major substations or
interconnecting systems, and to wholesale 1. Type of circuit Cable, overhead, single line,
outlets. Transmission lines are further divided parallel lines, multiterminals, etc.
into 2. Line function and importance Effect on service
continuity, realistic and practical time require-
High voltage (HV): 69 to 230 kV
ments to isolate the fault from the rest of the
Extra-high voltage (EHV): 345 to 765 kV
Ultra-high voltage (UHV): greater than 765 kV
3. Coordination and matching requirements
Compatibility with equipment on the associ-
ated lines and systems
4. Influence on power system stability
1.2 Techniques for Line Protection
To these four considerations must be added economic
Most faults experienced in a power system occur on factors and the relay engineer’s preferences based on
the lines connecting generating sources with usage his or her technical knowledge and experience. Because
points. Just as these circuits vary widely in their of these many considerations, it is not possible to
characteristics, configurations, length, and relative establish firm rules for line protection. This chapter,
importance, so do their protection and techniques. however, focuses on basic application rules and
There are several protective techniques commonly coordination procedures to aid the engineer in the
used for line protection: selection of proper protective systems for both phase

230 Chapter 12

and ground faults with the techniques as listed fact, these lines are the most difficult to protect,
in Section 1.2, except for pilot relaying. Also, this particularly with weak feed (limited-fault current) and
chapter covers the basic protective concept of series- when high-speed reclosing is desired at one or more
compensated transmission lines using distance relay terminals. Weak-feed protection may also be required
techniques. for two terminal lines. Although the weak-feed
Pilot relaying is covered in a companion book, Pilot terminal can maintain the fault arc, the current may
Protective Relaying (Marcel Dekker). not be sufficient to operate conventional protective
relays adequately.
The following definitions will be used throughout
1.4 Relays for Phase- and Ground-Fault this chapter:
Multiterminal lines. Transmission lines with more
Relay systems for the phase-fault protection of power than two terminals, each connected to a major
lines are outlined in Table 12-1, those for ground faults power source. The source will provide positive
in Table 12-2. sequence fault current and, usually, zero
sequence as well. A transformer bank may be
included as part of the transmission line at one or
1.5 Multiterminal and Tapped Lines and Weak more of the terminals.
Feed Tapped lines. Transmission lines that are tapped
(usually through a transformer bank) primarily
The protection of multiterminal and tapped lines and to supply loads. Behind the tapped line there may
weak feed will also be discussed in this chapter. be a positive sequence source, either local
Multiterminal and tapped lines, although usually generation or an interconnecting tie with another
economical in their breaker requirements, need com- part of the power system. There may also be a
plex relaying for adequate protection and operation. In zero sequece source.

Table 12-1 Relay Protection Systems for Phase Faults

Basic relay type

Type of protection no. Electromechanical Static or numericala

Time overcurrent 51 CO MCO, MMCO, 51

Instantaneous and time 50/51 CO with IIT MCO, MMCO, 51
overcurrent IMPRS, MICRO-51
Directional time overcurrent 67 CR 32 þ MMCO, 32 þ 51
32 þ MICRO-51
Directional instantaneous 67 KRV 32 þ 50Db
Step time overcurrent 51 CO-4 51 þ 50D
Directional instantaneous and 67 IRV 32 þ MMCO, 32 þ MICRO-51
directional time overcurrent
Inverse time distance system 21/51 Two KD-10, plus two-element CO
Zone distance system 21 Two KD-10, plus KD-11, plus
two TD-5 (or one TD-52)
Complete zone phase distance REL-300 (MDAR), REL-301/REL-302,
system REL-512
Complete distribution package DPU 1500R, DPU 2000R, MSOC
Type numbers refer to ABB circuit-shield types. Certain functions require two relays, with the output of the controlling relay wired to the torque-
control input of the second relay.
Select 50D with a 0.01 to 0.03 adjustable range.
Line and Circuit Protection 231

Table 12-2 Relay Protection for Ground Faults

Basic relay type

Type of protection Device no. Electromechanical Static or numericala

Time overcurrent 51N CO MSOC

Instantaneous and time overcurrent 50N/51N CO with IIT MSOC
Product overcurrent 67N CWC or CWP
Directional time overcurrent 67N CRC, CRP, CRD, or CRQ MSOC þ 32D or 32Q, MICRO-51
þ 32D or 32Q, 51 þ 32D or 32Q
Directional instantaneous 67N KRC, KRP, KRD, or KRQ 50D þ 32D or 32Qb
Directional instantaneous and time 67N/50N IRC, IRP, IRD, or IRQ MSOC þ 32D, 51 þ 32D, or 32Q
Complete zone ground distance REL-300 (MDAR), REL-301/REL-302,
system REL-512
Complete distribution package DPU 1500R, DPU 2000R
Type numbers refer to ABB circuit-shield types. Certain functions require two relays, with the output of the controlling relay wired to the
torque-control input of the second relay.
Select 50D with a 0.01 to 0.03 adjustable range.

Weak-feed terminal. A terminal whose source does tripped. To clear faults at (2) and (3) and in the area
not supply enough current for faults on the line between them, the breaker at H must be tripped.
to operate the line protective relays at that Likewise, to clear faults at (4) and (5) and between
terminal. This situation can occur for either them, the breaker at G must be tripped.
phase (positive sequence), ground (zero However, none of the relays at the breaker locations
sequence), or both. The terminals may only be can distinguish whether the remote fault is on the
‘‘weak’’ during some operating periods, but protected line, the remote bus, or an adjacent line. The
‘‘strong’’ or have only load at other times. A relays at H, for example, cannot distinguish between
tapped terminal is frequently a weak feed source faults at (1) and (2), since the current magnitude
if the tapped load has limited local generation, measured at H will be the same in either case. Opening
synchronous motors, and does not have many breaker H for fault (1) is not desirable, since it would
large induction motors. interrupt the load at R unnecessarily. Two techniques
are available to solve this problem: time delay or pilot
relaying. The latter requires a communication channel
2 OVERCURRENT PHASE- AND GROUND- between the two stations and is covered in the
FAULT PROTECTION companion book, Pilot Protective Relaying.

2.1 Fault Detection

Most of the faults on power lines can be detected by

applying overcurrent relays, since the fault currents are
normally higher than the load current.
Radial circuits can be protected by nondirectional
overcurrent relays. Figure 12-1 shows several sections
of a typical radial circuit. Because the circuit is radial,
each section requires only one circuit breaker at the
source end. To clear a fault at (1) and other faults to
the right, then, only the breaker at R needs to be Figure 12-1 A typical radial feeder.
232 Chapter 12

2.2 Time Overcurrent Protection

2.2.1 Time-Delay Relaying
Time relaying delays the operation of the relay for a
remote fault, allowing relays and breakers closer to the
fault to clear it, if possible. In the example shown in
Figure 12-1, relays at H will delay for faults at (1) or
(2). If the fault is at (1), this delay will allow the R
relays and breaker to operate before H. Thus, although
H would not open for a fault at (1) (unless the R relays
or associated breaker failed), it would operate for a
fault at (2). This technique, called coordination or
selectivity, is designed to combine minimum operating
time for the close-in faults with a long enough delay for
remote faults. In Figure 12-1, e.g., the relays and
breaker at R must coordinate or select with those to
the right (not shown), H must coordinate with R, and
G with H.

2.2.2 Coordination
Relays are coordinated in pairs. If, in Figure 12-1,
breaker H relay-tripping characteristics have already
been coordinated with whatever protective devices
exist at R and beyond, the breaker at G must then be
coordinated with those at H.
For the three critical fault points, (5), (3), and (2),
the following data are required:
1. Fault at (5). Maximum and minimum fault
2. Fault at (3). Maximum fault current, which
determines the required coordination between
breakers G and H Figure 12-2 Coordination setting procedure for relays at
3. Fault at (2). Minimum fault current, which breaker ‘‘G’’ of Figure 12-1.
determines when the G relays must operate to
provide backup protection for faults on line HR
not cleared by the breaker at H transformer ratio have been determined. (The selection
of these variables will be discussed later in this
Relays within a system can be coordinated using chapter.) Then perform the following:
graphs or tables, although graphs are generally more
useful for radial systems. Semilog (log abscissa for 1. Determine the critical fault locations and fault
current and linear ordinate for time) or log-log paper current values.
can be used. Log-log is preferred when a number of 2. Plot these variables on the time-current graph,
different types of devices, including fuses, are being drawing vertical lines at the various values.
coordinated on one graph. The current scale can be in 3. Determine the setting for the most downstream
primary amperes or per unit. Any difference in current relay for the maximum and minimum fault
transformer ratios must be taken into consideration currents. Set the relay as sensitive and fast as
when determining actual relay currents at different possible if there is no other device downstream
locations. that has to be coordinated with. For example,
The coordination procedure is conducted as follows consider the relay R in Figure 12-1. If there is
(Fig. 12-2). First, assume that the desired relay type some other device, such as a power fuse, to the
(tap range and time characteristic) and current right of this relay, then relay R should be
Line and Circuit Protection 233

coordinated with the power fuse first. If there 2.2.4 Selecting an Overcurrent Relay Tap
are no other devices to coordinate with down-
As indicated above, phase overcurrent relays must not
stream, set the overcurrent relay equal to or
operate on the maximum load current that can occur
greater than 2.0 maximum load.
on the line. Situations in which temporary overloads
4. Plot the operating time of relay R on the time-
may occur, such as the cold loads discussed in the next
current graph, shown as XR and YR points in
section, must be factored into the value used for setting
Figure 12-2, respectively.
the overcurrent relays. Thus, it is important for the
5. Add a one-step coordinating time interval
relay engineer to cooperate with the operating engi-
(CTI); (see Sec. 2.2.3) to points XR and YR.
neers in determining the maximum possible load for
This step gives two set points for the character-
each circuit. This maximum-value STM (short-time
istic curve of the relay at H.
maximum) load can differ from the rating of the line
6. Select a tap for relay H to operate for fault (1)
and is the value that should be used for setting the
minimum and, for a phase relay, not to operate
on maximum load. The fault (1) minimum
The tap (minimum pickup value of the phase
should operate the relay on at least twice
overcurrent relays) should be at least 2 (a safety factor
pickup, although compromises may be neces-
for security) times normal maximum load and never
sary (see Sec. 2.2.4). For phase relays, the
less than 1.5 times. If we assume that the STM is
setting must always be above the maximum
greater than the normal maximum load, the tap can be
selected as the next available tap greater than 1.25
7. Select a time lever such that the relay H time-
current curve passes through or above one or
Dependability should be checked once the relay tap
both of the set points XR and YR and provides
is selected. In Figure 12-8, the minimum fault current
the minimum operating time for maximum and
I2min through the relay for a fault on the remote bus H,
minimum fault.
divided by 2 (a dependability factor), should be greater
8. Repeat the above steps for each time section
than the selected tap value if the relay is set for
‘‘up-stream.’’ For example, add a one-step CTI
protecting the line only; or the minimum fault current
to XH and YH in Figure 12-2 for relay G,
I3min through the relay for a fault on the remote bus R,
respectively; then select a tap and time lever for
divided by 2, should be greater than the selected tap
relay G, etc.
value if the relay is set for protecting the line and
remote backup of the line (HR) beyond.
2.2.3 Coordinating Time Interval Current transformers are normally selected to
provide secondary currents between 4 and 5 A during
The coordinating time interval is the minimum time
rated maximum load. As a result, the phase relay
between the operating characteristics of two series
pickup will usually be above 5 A.
devices. Factors influencing the CTI are as follows:
The above limitation does not apply to ground
1. Breaker fault interruption time relays, since load current does not produce current in
2. Relay-impulse-time overtravel of the induction their operating windings unless it is unbalanced. To
disk or solid-state relay after the fault current avoid operation on possible imbalances in a normally
has been interrupted balanced circuit, a good rule of thumb is to set the
3. Safety margin to compensate for possible ground relays for not less than 10% of the maximum
deviations in calculated fault currents, relay load current. Four-wire distribution circuits will, in
tap selection, relay operating time, and current general, require a much higher setting than this.
transformer ratio errors
For coordinating at above approximately three times
2.2.5 Selecting an Overcurrent Relay Time
minimum trip current (at least two times the setting
value), the CTI should be in the range of 0.2 to 0.5 sec.
Larger CTIs should be used on the steep part of the Five different curve shapes have been established by
curve to compensate for errors below a multiplier of 3. the vast number of electromechanical relays that are in
A CTI of 0.3 sec is commonly used. Lower values service on power systems and that dictate coordination
should be used only after careful consideration of 1 requirements. Solid-state and numerical relays imple-
through 3 above. ment these shapes while often allowing others. These
234 Chapter 12

widely used time-current characteristics are described


1. Definite time, CO-6

2. Moderately inverse, CO-7
3. Inverse, CO-8
4. Very inverse, CO-9
5. Extremely inverse, CO-11

These time-current characteristics are compared in

Figure 12-3. The time lever settings are selected so that
all relays operate in 0.2 sec at 20 times the tap setting.
The microprocessor-based overcurrent relay, type
MCO, is a single-phase one, and type MMCO is a
three-phase and ground package. All the above time-
current curves are built in these relays, and can be
selected by settings. The equations for the MCO or
MMCO time-current curves are Figure 12-3 Type CO curve shape comparison.
TðsecÞ ¼ T0 þ for M  1:5 variations are available in the Micro-51, DPU 2000R,
ðM  CÞp 24; 000
  DPU 1500, and Microshield relays.
¼ for M < 1:5 The choice of a relay time-current characteristic is a
ðM  1Þ 24; 000 function of the sources, lines, and loads. Since these
factors vary throughout a system, a characteristic that
is ideal for one line and one operating condition
T ¼ trip time in seconds requires compromises for other conditions and associ-
D ¼ time dial setting from 1 to 63 ated lines.
M ¼ operating current in terms of multiple of tap If possible, time curves with the same or approxi-
setting mately the same characteristics should be used.
T0 ¼ definite time term Identical or similar curves applied at different places
K ¼ scale factor for the basic inverse time in the system tend to ‘‘track’’ together as operating
P ¼ an exponent determining inverseness conditions change. If different time characteristic
curves must be used, all possible operating conditions
T0, K, C, P, and R are constants and are shown as must be checked carefully to ensure that the CTI is
below: maintained for selective tripping. (Using similar
characteristics, in other words, minimizes coordination
Curve Fixed- and inverse-time characteristics for a system
no. T0 K C P R are compared in Figure 12-4. The last feeder supplying
one load center can be protected with an instantaneous
CO-2 111.99 735.00 0.675 1 501
CO-5 8196.67 13,768.94 1.130 1 22,705 overcurrent device set into the load. Since no
CO-6 784.52 671.01 1.190 1 1475 coordination at the load is involved, no time delay is
CO-7 524.84 3120.56 0.800 1 2491 required, as shown with bus R in both time-distance
CO-8 477.84 4122.08 1.270 1 9200 charts.
CO-9 310.01 2756.06 1.350 1 9342 In the upper chart of Figure 12-4, the fixed-time
CO-11 110.00 17,640.00 0.500 2 8875 characteristics approximate the definite minimum time
(CO-6). The relay at H is coordinated with R, and G
is coordinated with H, as shown. The advantage of
Operating time as shown in Table 12-3 illustrates this arrangement is that the operating times are
that the MCO or MMCO provides a fairly good time- relatively constant and independent of changes in
current characteristic for coordinating with the con- fault levels from maximum to minimum generation.
ventional type of CO relay. Similar curves and other On the other hand, the operating times for heavy
Line and Circuit Protection 235

Table 12-3 Relay Operation Times

Conventional CO MCO or MMCO

Time of operation Time of operation

Time dial at 4 6 pickup current Time dial at 4 6 pickup current
Type of relay set at (Fig. 12-3) (sec) set at (sec)

CO-6 0.6 0.25 6 0.25

CO-7 1.0 0.40 7 0.44
CO-8 1.2 0.60 7 0.58
CO-9 2.1 0.70 11 0.62
CO-11 5.0 2.00 31 2.00

faults near the source are very long. For this reason, produce very long operating times for minimum faults
this arrangement is not practical when there are more near the remote bus, it is commonly used.
than one or two radial feeders from the distribution Line length is also an important factor. For a short
substation. line, one whose impedance is low compared to the
The lower chart of Figure 12-4 shows inverse-time source impedance, the fault currents for the close-in
relay characteristics. For faults near the relay, parti- and far-end faults are essentially the same; that is, the
cularly for the maximum conditions, operating times inverse-time characteristic gives a relatively fixed
are very short. Unfortunately, as system conditions operating time over the line. In such cases, the definite
change from maximum to minimum, operating times minimum time characteristic is preferred, since the
vary considerably. Even though this arrangement can operating time will not vary as much for different
generation levels as with inverse relays.
In general, the following apply:

1. The flatter curves (CO-6 and CO-7) are more

suitable when:

(a) There are no coordination requirements

with other types of protection devices
farther out in the system.
(b) The variation in current for faults at the
near and far ends of the protected circuit is
too small to take advantage of the inverse
(c) Instantaneous trip units give good cover-
age (see Sec. 3.1).

2. The ‘‘inverse-time’’ relay (CO-8) provides faster

clearing time than the ‘‘more inverse-time’’
relays for low-current faults. This would be
advantageous on long lines where the available
fault current is much less at the end of the line
than at the local end. It does not provide much
margin for cold load pickup.
3. The steeper (more inverse) curves (CO-9 and
CO-11) are more suitable when

Figure 12-4 Comparison of fixed time vs. inverse time (a) Fault currents are significantly different
overcurrent relays on radial feeder circuits. for the close-in and remote faults (for
236 Chapter 12

example, when the line impedance is large

compared to the source impedance).
(b) There is an appreciable current inrush on
service restoration (cold load).
(c) Coordination with other types of devices
with very inverse characteristics, such as
fuses and reclosers, is required.

General Comments on Curve Shape Selection

There is no known scientific means of determining the
ideal curve shape for a specific application, except to
make preliminary setting calculations. However, some Figure 12-5 Typical example of feeder load current follow-
general comments can be made: ing extended outage (cold load inrush).

1. Use a CO-6 definite minimum time relay when maximum load, they will receive operating current
coordination is not a problem. for the first 2.3 sec that the feeder is energized (Fig. 12-
2. Use a CO-6 relay for short line application. 5). The average current for this period is 3.4 p.u., an
3. Use a CO-11, extreme inverse-time relay when equivalent of 1.7 times pickup. To prevent tripping the
fuses are involved. breaker, then, the relay operating time at 1.7 times the
4. The more inverse shape (CO-8, CO-9, and CO- tap value should be slightly more than 2.3 sec (about
11) is more suitable in loop systems. 2.5 sec). For the CO-9 very inverse relay, this condition
5. Use a comparable shape within a system requires a time dial setting of 1.25; for the CO-11
segment for easier coordination. extremely inverse relay, a time dial setting of 0.75 is
One of the advantages of using a microprocessor-based required. Time dial settings of 1.5 for the CO-9 relay
overcurrent relay is that a different time curve can be and 1.0 for the CO-11 relay are suggested, unless
selected in the same device without changing the unit. operating experience indicates otherwise.
Although the extremely inverse relays may provide
2.2.6 Effect of Extended Load Outage/Cold-Load faster fault operations than the less inverse type of
Inrush relays, they still override the cold-load inrush.
Modern microprocessor relays contain logic that
A particularly critical phenomenon for distribution allows sensing of the energization of a feeder circuit. A
circuits serving residential and commercial loads is the short time delay in instantaneous tripping can be
high transient current inrush that may occur when a introduced to prevent operation on cold-load pickup
feeder is energized after a prolonged outage. For these while still permitting sensitive and relatively fast fault
‘‘cold-load’’ conditions, the diversity of intermittent recognition.
loads is lost: Consumers tend to leave more than the Often, it is necessary to sectionalize the feeder and
normal load connected, and thermostatically con- to pick up the load in increments in order to reenergize
trolled equipment will start as soon as the voltage is the cold load without undesired tripping.
restored. The overall effect is a very high initial
current, or cold-load inrush.
2.2.7 Fuse and Relay Coordination
In general, the pickup of time-overcurrent relays
cannot be set above this transient without severely Because fuses have a time-current characteristic that is
compromising protection. Setting the relay below the much more inverse than most induction-disc time-
transient will cause it to begin to operate on the cold overcurrent characteristics, coordinating these relays
load; however, the current will decrease below the and fuses can be difficult (Fig. 12-6). A fuse curve and
pickup value before the relay has time to operate. two sets of relay curves, one for the extremely inverse
A current-time curve for the cold-load inrush on a relay and one for the very inverse relay, are plotted on
typical feeder is shown in Figure 12-5. Since such a linear time vs. logarithmic current scale. The right-
curves vary considerably with different feeders, each hand set of relay curves provides a good margin of
utility must develop its own system history and protection at high levels of fault current, but is
probability data. If we assume that the time-over- unsatisfactorily slow for medium values of fault
current CO relays are set at twice the normal current, particularly with extremely inverse character-
Line and Circuit Protection 237

Figure 12-7 Connections for overcurrent ground relay.

2.3 Instantaneous Overcurrent Protection

Adding instantaneous trip units to time-overcurrent

relays provides high-speed relay operation for close-in
faults and may also permit faster settings on the relays
in the adjacent section.
Instantaneous trips may be used on a circuit if the
Figure 12-6 Current (logarithmic scale) comparison of fuse
and relay curves.
maximum close-in fault current is on the order of 1.1
to 1.3 or more times the maximum fault current on bus
H (see Fig. 12-8).
istics. If the right curve were moved to the left, it would I1 max > ð1:1 to 1:3Þ I2 max ð12-1Þ
coordinate better at higher current values. As shown in
Figure 12-6, the curve would then cross the fuse curve The factor of (1.1 to 1.3) in Eq. (12-1) is for preventing
at lower values of fault current. Usually, either the very the instantaneous unit from overreaching. In other
inverse or extremely inverse characteristic can be set to words, the instantaneous unit must operate for as many
coordinate with fuses. By adjusting the tap and time of the line faults as possible, but to avoid miscoordina-
lever settings, areas of crossing such as those shown in tion, it must not operate for the far-end fault. The
Figure 12-6 may impose impractical or impossible greater the ratio of close-in to far-end faults, the more of
operating conditions on the circuit. the line the instantaneous unit will protect. In terms of
When plotting fuse curves, the following three time system constants and setting, the reach or coverage of
characteristics must be considered: three-phase faults on a line can be determined as follows:
SIRð1  Ki Þ þ 1
1. Maximum time that the fuse will carry current n¼ ð12-2Þ
without suffering damage Ki
2. Melting time for the fuse links
3. Total clearing time for the fuse to clear the

The first two characteristics are used for coordination

with protective devices beyond the fuse. Normally,
curves based on the melting-time characteristic are
provided with a ‘‘safety band’’; in this case,
maximum time curves are not required. The total
clearing-time characteristic is used for coordination
with other protective devices, including relays, ahead Figure 12-8 Criteria for a directional unit requirement at
of the fuse. relay breaker A.
238 Chapter 12

where their lower cost and complete independence of load,

the power system provides more rapid attenuation of
n ¼ per unit of line section length protected by
current with distance and the relatively higher inde-
the instantaneous unit
pendence of system changes. This makes their applica-
instantaneous unit pickup current; I
Ki ¼ maximum far-end fault current; I IT tion and setting easier than for phase relays.
SIR ¼ source impedance ratio Ground relays usually can be set and coordinated
source impedance; ZS independently of phase relays, even though the faulted
¼ protected line impedance; ZL phase current does flow through the one or more phase
relays for a single-phase-to-ground fault. The primary
Refer to Appendix A in this chapter for more reason for this independence is that ground relays are
information about Eq. (12-2). set at one-fifth to one-tenth of the sensitivity of phase
Recommended values of Ki, are 1.3 for the solenoid relays. The more sensitive settings obtainable with
or plunger units with transient overreach (IIT, SC, IT ground overcurrent relays may mean (for electro-
units), 1.2 for static or mumerical units (MCO, LI mechanical relays) a higher burden on the current
units), and 1.1 for the cylinder units with negligible transformers, and their performance should be
transient overreach (KC-2, KC-4, KO, KR, and IR checked as described in Chapter 5.
types). The value of 1.25 can be used as a general A circuit may be protected with a single, nondirec-
factor. tional overcurrent ground relay, as shown in Figure 12-
The minimum value that can justify the use of an 8. Positive and negative sequence currents are balanced
instantaneous unit for line protection is a matter of out at the current transformer neutral, so only 3I0
choice. Since the relative cost of adding the instanta- currents pass through the ground relay (50N/51N).
neous units is quite low, they are recommended even Since, under normal balanced conditions, 3I0, is at or
when the line coverage is low for maximum faults approaches 0, a very low pickup current is used,
and 0 for minimum faults. The arrangement provides typically 0.5 to 1.0 A. Although ground-fault currents
fast protection for the most severe, heavy, close-in on distribution circuits are generally higher at the
faults. substation than phase-fault currents, they decrease at a
Cold-load inrush may be above the instantaneous much greater rate with the distance from the sub-
unit setting desired for maximum fault protection. To station because X0 is considerably larger than X1 for
avoid operation when a setting above this inrush is the feeder circuits. With the exception of fault current
not practical, the instantaneous trip circuit can be values, the application and coordination of the
manually opened at restoration and left open until nondirectional overcurrent relays are the same as for
the instantaneous trip unit resets. For manual phase relays, as given above.
operation, a slip contact on the control switch that As with phase relays, instantaneous ground trip can
is open while the switch is held in the ‘‘close’’ be used to improve relaying, particularly for close-in
position (or its equivalent) prevents operation until faults. Instantaneous ground-trip units are more
the inrush subsides to drop out the instantaneous applicable in general, with the higher attenuation of
unit. If set above cold-load inrush, the instantaneous the fault currents with distance. Unless care is
trip unit setting should be at least three times the exercised in the choice of settings, high transient
overcurrent tap setting, or around six times the overloads and unequal current transformer perfor-
normal maximum load. mance can give rise to ‘‘false residual currents’’ and,
hence, misoperation.
The choice of a relay time characteristic for line
2.4 Overcurrent Ground-Fault Protection protection is usually limited to the inverse or very
inverse type. The very inverse type is the more
Ground overcurrent relays are for faults involving zero commonly used. However, when coordination with
sequence quantities, primarily single-phase-to-ground fuses and/or series trip reclosers is required, an
faults and sometimes two-phase-to-ground faults. extremely inverse characteristic would probably be
With a few significant differences, the general applica- preferable.
tion rules for phase relays also can be applied to The foregoing descriptions can be summarized as
ground relays. follows:
The directional or nondirectional overcurrent types 1. Factors that are favorable to ground-fault
are used widely at most voltage levels. In addition to protection:
Line and Circuit Protection 239

Normally load current does not affect the ground- procedure determines the criteria for a directional unit
relay operation. This means that the ground relay by comparing the currents flowing through the relay
can be set more sensitively than the phase relay. for faults at either bus.
Ground relays are not affected by out-of-step
conditions. 3.1.1 For Phase Directional Time-Overcurrent
Ground relays always have available the unfaulted Relays
phase voltages for polarizing. They may not
require a memory circuit. A directional time-overcurrent relay should be applied
The higher zero sequence line impedance Z0L, as at G if, as in Figure 12-8, the maximum reverse-fault
compared with the positive line impedance Z1L, current (I4max) for a fault on bus G or the maximum
may allow one to use a high-set ground over- reverse-load current IRLoad through the relay exceeds
current unit and make coordination easier than 0.25 (this value includes the safety factor 2 and
for phase faults. dependability factor 2) times the minimum fault
The zero sequence isolated system may make current I2min through the relay for a fault on the
coordination easier. remote bus H for the protection of the line only; or
the minimum I3min through the relay for a fault on the
2. Factors that are not favorable to ground-fault remote bus R if the relay is set for the protection of
protection: the line on the remote bus R and remote backup of the
Most of the time, ground faults involve higher fault line (HR) beyond. In other words, a directional relay
resistance; this may introduce an over- or under- should be used when
reach problem to the relay. I4 max or IRLoad
Zero sequence mutual effect may cause a ground- > ¼ 0:25 ð12-3Þ
I2 min or I3 min
relay directionality problem.
The zero sequence current distribution factor is not
equal to the positive sequence current distribu- 3.1.2 For Phase Directional Instantaneous-Trip
tion factor, except for a single end feed condition; Overcurrent Relays
this makes the ground relay more complicated
than the phase relay in design. A directional instantaneous-trip overcurrent unit
The ground relay faces more problems than the should be used if the maximum reverse-fault current
phase relay on reverse fault clearing (e.g., contact I4max is greater than the maximum I2max (Fig. 12-8).
bounce, unequal pole clearing, etc.). Also, as mentioned in Section 2.3 before, for instanta-
neous-trip overcurrent application, the criteria as
shown in Eq. (12-1) should be met.


GROUND-FAULT PROTECTION 3.2 Criteria for Ground Directional
Overcurrent Relay Applications
3.1 Criteria for Phase Directional Overcurrent
Relay Applications With a few significant differences, the general rules of
application for directional overcurrent phase relays
When there is a source at more than one of the line also apply to ground relays. Normal balanced load
terminals, fault and load current can flow in either current is not a consideration. However, the ground
direction. Relays protecting the line are therefore overcurrent unit still should be set above any
subject to fault power and reactive flowing in both maximum expected unbalanced load current.
directions. If nondirectional relays were used, they
would have to be coordinated with not only relays at
the remote end of the line, but also the relays behind 3.3 Directional Ground-Relay Polarization
them. Since directional relays operate only when fault
current flows in the specified tripping direction, they To determine the direction to a fault, a directional
avoid both this complex coordination and the possi- relay requires a reference against which line current
bility of compromising line protection. can be compared. This reference is known as the
Figure 12-8 shows a line, with a source at each end, polarizing quantity and, in this context, reference and
that could be a section of a loop. The following polarizing are synonymous terms. With zero sequence
240 Chapter 12

line current, either a zero sequence current or voltage 3.3.2 Current Polarization
or both must be used. In power systems with mutual
The current polarization reference depends on the
induction problems, the trend is toward the use of
availability and connection of the power transformer
negative sequence quantities for the ground directional
at the relay location. Various bank connections are
diagrammed in Figure 12-10. The current in the
neutral of a wye-grounded delta power transformer
3.3.1 Voltage (Potential) Polarization
can be used for polarizing. For almost any ground
The zero sequence voltage at or near any bus in an fault on the wye-side system, this current flows up the
inductive power system can be used for polarization. neutral when current is flowing to the fault. (An
The voltage measured on the bus or just on any line exception with high mutual induction will be discussed
near the station will have the same direction for any later.) Thus, a current transformer in the neutral
fault location. However, the current through the line measures 3I0 for IP current polarization, the polarizing
breaker will change direction according to the fault current shown in Figure 12-10a for a wye-delta bank,
location. and that shown in Figure 12-10d for a zig-zag bank.
The polarizing zero sequence voltage is obtained Wye-wye banks, either grounded or ungrounded,
from the broken-delta secondary of grounded-wye cannot be used for polarizing (Fig. 12-10b and
voltage transformers (Fig. 12-9). Phase voltages are 12-10c). The grounded wye of three-winding
also required for the phase relays, instrumentation, etc. wye-wye-delta banks can be used (Fig. 12-10e, f,
In such cases, either a double-secondary voltage and g).
transformer or device, or a set of auxiliary wye- The separate neutral currents of the three-winding
grounded, broken-delta auxiliary transformers, can be wye-delta-wye transformers cannot be used for polar-
used. The voltage across the broken delta, VXY, always izing, but current transformers in each grounded
equals 3 V0, or VAG plus VBG plus VCG. neutral must be parallel with inverse ratios, as shown
in Figures 12-10f and 12-10g. If we assume that both
the high- and low-voltage sides connect to a ground
source, ground faults on the low-voltage side (Fig. 12-
10g) result in current flowing up the low-voltage
neutral and down the high-voltage neutral. Conversely,
for faults on the high-voltage side (Fig. 12-10f), current
flows up the high-voltage neutral and down the low-
voltage neutral. Hence, the reversal in either neutral
does not provide a reference. By paralleling the two
neutral current transformers, however, IP always falls
in the same direction for faults on either side since, on
a per unit basis, the current flowing down the neutral is
always less than the current flowing up the other
neutral. The actual current distribution will vary as
determined by the zero sequence network.
The tertiary or delta winding can also be used as a
polarizing source. If there are no external circuits from
the delta, one current transformer connected in any leg
of the delta will provide I0. A current transformer is
required in each of the three windings if the delta is
connected to external circuits, so that positive and/or
negative sequence currents can exist during load or
faults. These current transformers must be connected
in parallel to cancel out positive and negative sequence
and provide 3I0 only.
Autotransformers should not be used for current
polarizing without careful analysis because they are
Figure 12-9 Zero sequence polarizing voltage source. frequently unreliable as a reference. Autotransformers
Line and Circuit Protection 241

Figure 12-10 Zero sequence polarization from power transformer banks.

that are ungrounded or without a delta tertiary cannot of the bank is

be used. Such units can bypass zero sequence in  
roughly the same way as the wye-wye-grounded banks. VH
INH ¼ 3 I0H  P0 I0H
A grounded autotransformer with a tertiary is shown VL
in Figure 12-11.  
¼ 3I0H 1  P0 ð12-4Þ

For Ground Faults on the High-Voltage Side From this, it is evident that a positive value of ð1 
(Figure 12-11a, if we assume the high-side source is P0 VH =VL Þ will assure that the neutral is a reliable
opened for simplicity). The zero sequence current polarizing source.
flowing through the high-voltage winding of the From the zero sequence network, we obtain
autotransformer is thus I0H. The current flowing in  
the low-voltage circuit will be P0I0H(VH/VL). Because ZT
P0 ¼ ð12-5Þ
of the physical connections, the current in the neutral Z0LS þ ZL þ ZT
242 Chapter 12

Figure 12-11 Polarizing from grounded autotransformer banks.

Substituting yields from the high voltage system will be R0I0L(VL/VH) at

  VH. Again, the bank neutral current is
INH ¼ 3I0H 6 1  6 ð12-6Þ  
INH ¼ 3 I0L  R0 I0L
The tertiary current (inside the delta) is VH
VH ¼ 3I0L 1  R0 ð12-10Þ
ITH ¼ I0H ð1  P0 Þ pffiffiffi at VT ð12-7Þ VH
3V T
and since From the zero sequence network, we get
ð1  P0 Þ ¼ ð12-8Þ R0 ¼ ð12-11Þ
Z0SL þ ZL þ ZT Z0SH þ ZH þ ZT
ITH ¼ I0H Substituting yields
Z0SL þ ZL þ ZT
6 pffiffiffi at VT ð12-9Þ ZT VL
3V T INL ¼ 3I0L 6 1  6

For Ground Faults on the Low Voltage Side The tertiary current (inside the delta) is
(Figure 12-11b, if we assume the low-side source is
opened for simplicity). I0L is the zero sequence current ITL ¼ I0L ð1  R0 Þ pffiffiffi at VT ð12-13Þ
in amperes from the autotransformer. The current 3VT
Line and Circuit Protection 243

and since less than 10% (usually they do not exceed 3 to 7%) of
  self-impedances and can be considered negligible. For
ð1  R0 Þ ¼ ð12-14Þ zero sequence, however, mutual impedance can be 50
Z0SH þ ZH þ ZT to 70% of the zero sequence self-impedance Z0L and is,
Z0SH þ ZH therefore, significant. Mutual impedance affects the
Z0SH þ ZH þ ZT magnitude of ground-fault currents and can result in
VL incorrect directional sensing.
6 pffiffiffi at VT ð12-15Þ Figure 12-12 shows two parallel three-phase lines
3V T
with zero sequence isolation, except for mutual
If the neutral or tertiary current is used for coupling. The two lines can be completely isolated,
polarizing, Eq. (12-6) and (12-12) or (12-9) and (12- or more commonly, tied together to common genera-
15) must give a positive operating value for all possible tion sources. The lines can be at the same or different
variations of Z0SH and Z0SL. However, in some voltage levels.
transformer designs, especially those for autotransfor- A ground fault at or near G on line GH involves no
mers, the ZH or ZL may be negative, the direction of directional sensing problems at either station G or H,
the INH or INL may be reversed, and this current is not as shown by the I0 current directional arrows. The 3I0
a reliable source for polarization, when the combined current flowing from H to G induces a zero sequence
impedance of (Z0SH þ ZH) or (Z0SL þ ZL) is a negative voltage in the parallel line RS, causing current to flow
value. from R to S as shown. The polarizing and operating
quantities are properly oriented to operate the zero
3.3.3 Dual Polarization sequence quantity-polarized directional ground relays
at both R and S, so that this current appears as an
The approach is called dual polarization if the internal line RS fault to the relays at terminals R and
directional ground overcurrent relay uses both zero S.
sequence voltage 3V0 and zero sequence current Ip for The zero sequence current I0M in line RS induced by
polarization. It provides more flexibility in applica- uniform mutual coupling (Z0M) for length GH will be
K0 I0 ðnZ0M Þ þ ð1  k0 ÞI0 ð1  nÞZ0M
I0M ¼
3.3.4 Negative Sequence Polarization Z0SR þ Z0L þ Z0SS
½1  ðn þ K0 ÞI0 Z0M
The negative sequence directional ground unit is ¼
operated by the quantities V2 and I2. One typical Z0SR þ Z0L þ Z0SS
design operates when I2 leads V2 by 988. The output of DVSR
¼ ð12-16Þ
the negative sequence directional unit D2 can be used Z0SR þ Z0L þ Z0SS
for either supervising or the torque control of an I0 or
I2 unit. where
Negative sequence relays can be tested easily and n ¼ per unit fraction of Z00L from bus G to the
quickly using load current flow. The directional unit is fault
checked by simply interchanging B and C currents and DVSR ¼ induced voltage drop from S to R
voltages to provide negative sequence from the
balanced load quantities. A